single-cell chemical lysis method for analyses of … · single-cell chemical lysis method for...

Anal. Chem. XXXX, xxx, 000-000

Single-Cell Chemical Lysis Method for Analyses of lntracellular Molecules Using an Array of Picoliteu-Scale Microwells

Yasuhiro Sasuga,'$* Tomoyuki I w a s a ~ a , ~ Kayoko Terada,'** Yoshihiro Oe,'l5 Hiroyuki S~uimachi ,~ Osamu Ohara,*v"9L and Yoshie Haradat'*rsyl"

The Tokyo Metropolitan Institute of Medical Science, 3- 18-22 Honkomagome, Bunkyo-lcu, Tokyo 1 13-86 13, Japan, Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency (JST), Honcho, Kawaguchi, Seitama 332-0012, Japan, Department of Medical Genome Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Building FSB-401, 5-1-5, Kashiwanoha, Kashiwa, Chiba 277-8562, Japan, Depaiiment of Human Genome Research, Kazusa DNA Research Institute, 2-6-7 Kazusa-kamatarl, Kisarazu, Chiba 292-0818, Japan, Research Center for Allergy and immunology, RlKEN Yokohama lnsfitute, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa 230-0045, Japan, and Institute for Integrated Cell-Material Sciences (iCeMS), Kyoto University, Yoshida-konoecho 69, Sakyo-ku, Kyoto 6068501, Japan

Analyzing the illtracellular contents and enzymatic activities of single cells is importailt for studying the physiological and pathological activities at the cellular level, For this purpose, we developed a simple single-cell lysis method by using a dense array of microwells of 10-30- pL volume fabricated by poly(diinethylsi1oxane) (PDMS) and a coinmercially available cell lysis reagent. To demonstrate the perforinance of this single-cell lysis method, we carried out two different assays at the single-cell level: detection of proteins by antibody conjugated microbeads and measurement of protease activity by fluorescent substrates. The results indicated that this method readily enabled us to inonitor protein levels and ei~zymatic activities in a single cell. Because this method required only an array of PDMS microwells and a fl~iorescence microscope, the simplicity of this platfonn opens a way to explore the biochemical characteristics of single cells even by those who are not familiar with microfluidic technology,

A cell is the fundamental unit of life, and all lunctions in ~l~ullicelltrlar organisms are ultiniately attri1)ut:ecl to those of the cell. Although most conve~~tio~~al biochemical assays are per- lorlned using a considerable niunber of cells to determine their quantitative I~iomolecular profiles, such bulk assays only provide their averaged values in the analyzed ensemble and thus oRe11 oveiAloolc imporlant inforlnalion regarding their flttc(ua1ions anlong individual ~el ls . l*~ Becatrse the dilferences anlong individual cells in the saine ense~nble are highly critical in sonle cases such as

*To whom co~.respondence should be aildressed. IJlione: +81-4%-52-3913. Fax: -1-81-438-52-3914. Emall: oliarnQltazusa,or.jp.

IThc 'Soltyo Mc(roj)olitan Inslitulc ol Mcdicai Scicncc. * CRRSS, JST. 3 Tile University oi Toltyo. " ICazusa DNA I<esenrch Instilute. .'- IIIIiI<N Yoltohami~ Instilutc. ' Iiyoto U~iiv~rsity,

(1) I;er~.ells, J. E.; Machledel; E, M. Sciet~cc 1998, 280, 8'35-898, (2) Levsky, J. M.; Singer, R. H. T~eiuis Cell Uiol. 2003, 13, 4-6,

10.1021/ac8016423 CCC: $40.75 O XXXX American Chemical Society Published on Web 11/05/2008

cell diflerentialion, quantitative measure~nent of biolnolecular profiles at the single-cell level is a matter of concern in the biological community. Should it be possible to determine the biochemical parameters of each of multiple single cells in parallel, we would be able to describe the behavior of individual cells in an ensemble ior obtaining deeper insights into cell-cell signaling, genetic helerogeneity, and lleterotypic biological activiliese3 Furlhemore, if single-cell biochemical analysis becomes feasible, we will be able to save much time, labor, and cost because we no longer have lo purify a cell ensen~ble lo homogeneity.

To develop a single-cell analysis in quantitative biology, various methods have been actively e~plored.~-~owever, these newly emerging methods have not been fi~lly applied in the biological scieilces yet. One of the reasons lor this is the fact that these n~elhods are too sopl~isticated and integrated to be used without appropriate inveslment oi time, money, and labor. Thus, there is a strong ilecd to simplify anrl ~nalte these methods more biologisl- friendly for realization of quantitative biology at the single-cell level, We believed that the difflculty lies in Lhe [act that quantitative analysis of intracellular biological contents at the single-cell level must seatnlessly integrate multiple steps: cell isolation/lrapping, cell lysis, and quanlilative assays of biochelnical contents in the lysate, As lor ~nultiplexed single-cell analyses, most inethods reported thus far take advantage of highly sophisticated and automated instru~nents for integration of these multiple steps on a single platfor~n, eg,, single-cell caplure followed by chen~ical lysis in a closed volume of 50 pL recently reported in a micro-

(3) Rubin, M. A. Scbt~cc 2002, 296, 1329-1330. (4) Levslry, J. M.; Slienoy, S. M.; I'ezo, R. C.; Singer, R. H. Scietice 2002,297,

833-840. (5) Iioog, J. W,; SLudcr, V.; Hang, G.; Antlcrson, W. I?.; Qualtc, St 11. Not.

Biot~cktrul. 2004, 22, 435-439. ((i) I<urimoto, IC.; Yabuta, Y.; Ohinata, Y.; Olio, Y.; Uno, I<. I).; Ynmacla, It. C;.;

Ueda, I-I. R.; Sailou, M. Nzicleic Acids Res. 2006, 34, e42. (7) Huang, 13.; Wu, H.; Nht~ya, D.; C;rossman, A,; Granicr, St; I<obillti~, 13. I<.;

Zarcl, I<. N. Sciancc 2007, 31.5, 81-84, (8) Newman, J. S.; Ghaann~eghami, S.; Il~nlels, J.: Breslow, D. I<.; Noble, M.;

DeIGsi, J. L.; Weiss~nan, J. S. Nntztre 2006, 441, 840-846.

Analytical Chemistry, Vol. xxx, No. xx, Month XX, XXXX A

Biophysical Journal Volume 95 October 2008 3429-3438 3429

Monolayers of Silver Nanoparticles Decrease Photobleaching: Application to Muscle Myofibrils

P. Muthu,* N. Calander," I . ~ r y c z y n s k i , ~ Z. Gryczynski,* J, M. Talent," T. ~ h t o y k o , ~ I , Akopova," and J. Borejdo* *Department of Molecular Biology and Immunology, and +~epartrnent of Cell Biology and Genetics, University of North Texas Health Science Center, Fort Worth, Texas; and *~epartment of Chemistry, University of Texas at Tyler, Tyler, Texas

ABSTRACT Studying single molecules in a cell has the essential advantage that kinetic information is not averaged out. However, since fluorescence is faint, such studies require that the sample be illuminated with the intense light beam. This causes photodamage of labeled proteins and rapid photobleaching of thefluorophores. Here, we show that a substantial reduction of these types of photodamage can be achieved by imaging samples on coverslips coated with monolayers of silver nanoparticles. The mechanism responsible forthis effect is the interaction of localized surface plasmon polaritons excited In the metallic nanoparticles with the transition dipoles of fluorophores of a sample. This leads to a significant enhancement of fluorescence and a decrease of fluorescence lifetime of a fluorophore. Enhancement of fluorescence leads to the reduction of photodamage, because the sample can be illuminated with a dim light, and decrease of fluorescence lifetime leads to reduction of photobleaching because the fluorophore spends less time in the excited state, where it is susceptible to oxygen attack, Fluorescence enhancement and reduction of photobleaching on rough metallic surfaces are usually accompanied by a loss of optical resolution due to refraction of light by particles. In the case of monolayers of silver nanoparticles, however, the surface is smooth and glossy. The fluorescence enhancement and the reduction of photobleaching are achieved without sacrificing the optical resolution of a microscope. Skeletal muscle myofibrils were used a s an example, because they contain submicron structures conveniently used to define optical resolution. Small nanoparticles (diameter -60 nm) did not cause loss of optical resolution, and they enhanced fluorescence -500-fold and caused the appearance of a major picosecond component of lifetime decay. As a result, the sample photobleached -20-fold more slowly than the sample on glass coverslips.

INTRODUCTION

Recently, it has become possible to study single protein rophore, and therefore has no effect on the intrinsic resistance molecules in a cell (1,2). The advantage of single-molecule to photobleaching, it increases the time that a fluorophore detection (SMD) is that it studies the behavior of proteins in remains unbleached, because the sample can be illuminated their native (crowded) environment, and avoids problems with a weaker light. Decrease of fluorescence lifetime, on the associated with averaging responses of an assembly of mol- other hand, leads to a decrease of the intrinsic photobleaching ecules with different kinetics. However, the single-molecule rate. The decrease in bleaching is mainly caused by the fact

1 approach is complicated by photobleaching, which arises that the molecule spends less time in the excited state, where 1 , I J - " (

because the sample must be illuminated with the intense laser it is subject to oxygen attack. For a fluorophore in a normal \ 1 (. beam to assure an adequate signal/noise ratio. nonsaturation limit (i.e., the increase in laser power causes a

In an earlier work, we were able to decrease photo- corresponding increase in fluorescence intensity), the de- bleaching by making measurements on coverglasses coated crease of fluorescence lifetime also causes an increase in with nanoparticles known as surface island films (SIF) (3). quantum yield, Q. However, since the Q of rhodamine is near SIFs are nanoparticles (4-10) that can support confined 1 anyway, this effect is not very significant in our experiments. charge density oscillations (1 1) called localized surface For a fluorophore in a saturation limit (i.e., the increase in plasmon (LSP) polariton modes. Excitation of LSPs causes a laser power does not cause an increase in fluorescence in- strong ellhancement of local electric field and a substantial tensity), there is an additional effect: the decrease in fluo- decrease of fluorescence lifetime caused by the distance- rescence lifetime causes an increase of effective cross section dependent changes in the radiative decay rates (12). Enhance- for absorption; because short-lived fluorophores can now be ment of the local field allows attenuation of illumination, excited, whereas long-lived ones cannot be, which leads to reduction of photodamage. At the same time, As a consequence of SF-induced field enhancement and enhancement causes an effective decrease of photobleaching: decrease of lifetime, the rate of photobleaching of rhoda- although it does not change the characteristics of a fluo- mine-phalloidin-labeled actin in a myofibril placed on glass

coverslips coated with SIF decreased approximately twofold - - Subnlittcd February 4, 2008, and accepted for. p~rblicatiorl April 7, 2008,

in comparison with that of myofibrils placed on uncoated

Address reprint requests to Julian Borejdo, Dept, of Biochetnistry, Health coverslips (3). The price of the reduction in bleaching,

Science Center, University of North Texas, 3500 Camp Bowie Blvd., Fort however, was a loss of 'ptical resolutioll and loss of Worth, TX 76107-2699. Tel,: 817-735-2106; E-~~lail: jborejdo@hsc,unt, uniformity of illumination, because SIF refracted the exciting edu. and fluorescent light, and because nanoparticles were dis- Editor: Elliot L, Elson. O 2008 by the Biophysical Society 0006-3495/08/10/3429/10 $2.00 doi: 10.1529/biop11ysj.108~130799

Leading Edge

A Social Life for Discerning Microbes Sam P. Brown1#* and Angus Buckling1 'Department of Zoology, University of Oxford, Oxford OX1 3PS, UK *Correspondence: [email protected] DO1 10.1016/j.ce11.2008.10.030

Microbes are not only extremely social but also extremely discerning about whom they socialize with. Recent research has uncovered some of the evolutionary explanations behind these feats of social sophistication in bacteria (Ackermann et al., 2008; Diggle et al., 2007) and, most recently, has provided insights into the molecular mechanisms of discrimination in yeast (Smukalla et al., 2008).

Cooperative acts ranging from minor help to marked self-sacrifice are seen in all corners of biology, from microbes to man. Given the presence of "freeloading cheaters," why is it that cooperative behaviors persist? Following the pioneering work of Hamilton (1964), a consensus has emerged that cooperative behaviors evolve and are maintained through a mix of self- interest and nepotism, which maximizes an individual's inclusive fitness, that is, the reproductive success of an individual and its close relatives. Cooperation may be self-interested if it directly benefits the actor as well as the recipients (for instance, increasing the success of one's own group). More extreme forms of cooperation-for example, altruistic cooperation where individuals experience a direct fitness cost in helping others-may be favored because the behavior helps recipients who are likely to share the altruistic gene (Hamilton, 1964). In other words, natural selection will cause a gene to spread if it confers an advantage on the individual in which the gene is present, as well as if it confers an advantage on other individuals with the same gene. But for this latter process to operate, altruistic acts must be preferentially directed toward other altruists. The most common scenario is that altruism is expressed blindly to neighbors, who will tend to be relatives (with an overrepresentation of similar genes) due simply to population viscosity (there is not complete mixing of individuals within the population) (Hamilton, 1964). A more complex scenario involves active processes of kin recognition and discrimination. However, this is not a perfect system as kin may still be "cheaters" that lack the vital gene for altruism.

Hamilton (1964) realized that one foolproof way to avoid wasting help on cheaters is for an individual to clearly display an altruistic or social gene and to recognize directly the same gene in others, rather than relying on proxy identifiers such as location or kinship. An individual social gene displays itself and enables recognition of the same gene in others, thus ensuring that help is only conferred on individuals expressing that gene. Richard Dawkins (1976) popularized this notion as the "green beard" gene: a gene that can be recognized externally because it confers a green beard on its carrier, ensuring that only green- bearded individuals are helped.

Hunting for Green Beard Genes Green beard genes remained a plausible thought experiment until a remarkable empirical example was reported by Keller and Ross (1998): They discovered a gene cluster linked to the

display and discrimination of identity in red fireants. Next, Quel- ler et al. (2003) identified an even tighter association between display, discrimination, and social traits in the first clear docu- mentation of a single green beard gene in the slime mold Dic- tyostelium discoideum. When food is plentiful, Dictyostelium amoebae lead a single-celled, individualistic life of consump- tion and (asexual) reproduction. However, once starvation arrives they aggregate to form multicellular assemblies, replete with complex signaling mechanisms, division of labor, and individual sacrifice. The multicellular slugs differentiate into a sacrificial stalk, on top of which sit the lucky spores, which can then hitch a ride on invertebrates to potentially more favorable destinations. Queller et al. (2003) demonstrated that a single gene is responsible for holding nonstalk-forming "cheaters" at bay, by controlling entry to the spores. The csaA gene encodes a cell adhesion protein anchored in the cell membrane (the discernable green beard), which binds to homologous adhesion proteins (discrimination), building a multicellular aggregate preferentially of csaA carriers (targeted cooperation). This simple mechanism highlights the ability of microbes to harness molecular tricks to build sophisticated social behaviors (Fos- ter et al., 2007), raising the possibility that these feats may be more common in microbes than previously suspected. Indeed, the bacterium Proteus mirabilis will only form motile swarms with those of the same strain. Although the molecular mechanism behind this recognition has not yet been fully deduced, a set of genes required for the recognition of "like" cells has been identified in P. mirabilis (Gibbs et al., 2008).

A Discriminating Yeast The latest addition to the green beard gene stable, as Smukalla et al. (2008) report in this issue of Cell, appears in an unex- pected microbe, the budding yeast Saccharomyces cerevisiae. Most strains of S. cerevisiae display an aggregating response to environmental stress called flocculation, Smukalla et al. (2008) now reveal that the expression of a single gene, FLOI, restores flocculation to a laboratory strain of S. cerevisiae, S288C, generating, at an individual cost, a social protection for S288C yeast within the aggregate (floc) against diverse environmental stresses such as ethanol and fungicides. But what about the potential for nonexpressing fiol cells to act as cheai- ers, exploiting the protection of established aggregates without paying the cost associated with FLOI expression? Smukalla et

nor\ rr.8, * n r r,.. 8 . 4 r nnnn mnnno r,- .. :... ,..

Molecular Cell

Molecular Basis for Regulation of the Heat Shock Transcription Factor c3* by the DnaK and DnaJ chaperones Fernanda R o d r i g ~ e z , ~ ~ ~ Florence Arskne-Ploet~e,' ,~ Wolfgang Rist,lp5 Stefan Riidiger,lfjJens Schneider-Mergener,' Matthias P. Mayer,'l* and Bernd Bukaul~* 'Zentrum fiir Molekulare Biologie der Universitat Heidelberg, DKFZ-ZMBH Alliance, Im Neuenheimer Feld 282, 69120 Heidelberg, Germany *lnstitut fur Medizinische Immunologie, Un~versitatsklinikum Chant&, Schumannstrasse 20-21, 10098 Berlin, Germany, and the Jerini AG, lnvalidenstrasse 130, 101 15 Berl~n, Germany 3Present address: Growth and Development, Blozentrum, University of Basel, Klingelbergstrasse 50170, CH-4056 Basel, Switzerland 4Present address: UMR7156, UniversrtB Louis PasteurICNRS, Laboratoire de GBnetique MolBculaire, Genomique et Microbiologie, Departement Micro-organismes, GBnomes, Environnement, 28, rue Goethe, 67083 Strasbourg Cedex, France 5Present address: Boehringer lngelheim Pharma GmbH, Department of Respiratory Research, Proteomics Laboratory, Birkendorfer Strasse 65,88397 B~berach, Germany 6Present address: Cellular Protein Chemistry, Department of Chemistry, Utrecht University, Kruytgebouw, Room 0-701, Padualaan 8, 3584CH, Utrecht, the Netherlands "Correspondence: m.mayerOzmbh.uni-heideiberg.de (M.P.M.), bukauOzmbh.uni-heideIberg.de (B.B.) DO1 10.1 01 6/j.molce1.2008.09.016

SUMMARY

Central to the transcriptional control of the Escheri- chia coli heat shock regulon is the stress-dependent inhibition of the 032 subunit of RNA polymerase by reversible association with the DnaK chaperone, mediated by the DnaJ cochaperone. Here we identified two distinct sites in 032 as binding sites for DnaK and DnaJ. DnaJ binding destabilizes a distant region of 032 in close spatial vicinity of the DnaK-binding site, and DnaK destabilizes a region in the N-terminal domain, the primary target for the FtsH protease, which degrades 032 in vivo. Our findings suggest a molecular mechanism for the DnaK- and DnaJ- mediated inactivation of 032 as part of the heat shock response. They furthermore demonstrate that DnaK and DnaJ binding can induce conformational changes in a native protein substrate even at distant sites, a feature that we propose to be of general relevance for the action of Hsp70 chaperone systems.

INTRODUCTION

The heat shock response is an evolutionary conserved protective mechanism of cells against stress-induced damage of proteins. In E. coli, this response is mediated by the rpoH gene product, the heat shock transcription factor 03', that binds as an alternative cr subunit to the RNA polymerase (RNAP) core enzyme and targets it to the promoters of heat shock genes (Bukau, 1993; Gross, 1996; Yura and Nakahigashi, 1999). Stress-dependent changes in heat shock gene expression are mediated by changes in the activity and stability of 03'. In cells growing at 30°C, 03' is rapidly degraded (tl12 i 1 min) primarily by the membrane-bound

AAA-protease FtsH, which accounts for its very low intracellular levels (10-30 molecules per cell). Upon temperature upshift from 30°C to 42"C, the levels and half-life of 03' increase transiently (induction phase). The induction phase is followed by a shut-off phase during which the synthesis of heat shock proteins is reduced to a new steady-state level. This shut-off results from rapid inactivation and destabilization of 03' (Herman et al., 1995; Kanemori et al., 1999; Straus et al., 1987; Tilly et al., ,1989; Tomoyasu et al., 1995).

Genetic and biochemical ,evidences indicate that the major Hsp70 chaperone in E. coli, DnaK, and its cochaperones DnaJ and GrpE play acentral role forthe adjustment of thesteady-state levels of heat shock proteins and the rapid shut-off of the heat shock response (Grossman et al., 1987; Straus et al., 1990; Tilly et al., 1983). In vitro, 03' in its native state is recognized as a substrate by DnaK and its cochaperone DnaJ (Gamer et al., 1992, 1996; Liberek et al., 1992, 1995). The interaction of DnaK with 03' is transient and controlled by DnaK's nucleotide status. In the ADPstate, the affinity of DnaKto 03' is high and association and dissociation rates are low, while in the ATP state the affinity is low and association and dissoclatlon rates are high (Gamer et al., 1996; Liberek et al., 1995; Mayer et al., 2000). DnaJ binds 03' with high affinity, and simultaneous interaction of DnaJ and 03' with DnaK#ATP stimulates DnaK's ATPase activity several thou- sand-fold, leading to the tight binding of DnaK+ADP to cr3' (Gameret al., 1992,1996; Laufen et al., 1999; Libereket al., 1995).

The precise mechanism by which DnaK and DnaJ interact with 03' and regulate its activity and stability is still unclear. By screening libraries of peptides derived from many different chaperone substrates, earlier work established the consensus motif that DnaKand DnaJ recognize in substrates. For DnaK It Is composed of a core of five amino acids enriched in hydrophobic residues flanked by sequences enriched in positively charged amino acids (Riidiger et al., 1997). By screening a 03'-derived peptide library for DnaK-binding sites, seven sites were identified within the 03'

Cytokine-Induced Signaling Networks Prioritize Dynamic Range over Signal Strength Kevin A. J a n e ~ , ~ ~ ~ ~ ~ ~ ~ H. Christian Reinhardt,lp3 and Michael B. Yaffe11* 'Koch lnstitute for Integrative Cancer Research, Center for Cell Decision Processes, Department of Biology, and Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 2Department of Cell Biology, Harvard Medical School, Boston, MA 02115, USA 3These authors contributed equally to this work 4Present address: Department of Biomedical Engineering, University of Virginia, Charlottesville, VA 22908, USA "Correspondence: myaffeQmit.edu DO1 10.1 01 6/j.ce11.2008.08.034

SUMMARY

Signaling networks respond to diverse stimuli, but how the state of the signaling network is relayed to downstream cellular responses is unclear. We modeled how incremental activation of signaling molecules is transmitted to control apoptosis as a function of signal strength and dynamic range. A linear relationship between signal input and response output, with the dynamic range of signaling moleci~les uniformly distributed across activation states, most accurately predicted cellc~lar responses. When nonli- nearized signals with compressed dynamic range relay network activation to apoptosis, we observe catastrophic, stimulus-specific predictiori failures. We develop a general computational technique, "model-breakpoint analysis," to analyze the mechanism of these failures, identifying new time- and stimulus-specific roles forAkt, ERK, and MK2 kinase activity in apoptosis, which were experimentally verified. Dynamic range is rarely measured in signal-transduc- tion studies, but our experiments using model-breakpoint analysis suggest it may be a greater determinant of cell fate than measured signal strength.

INTRODUCTION

Changes in cell behavior are determined by an interconnected set of proteins that actively transmit signaling information as a network (Irish et al., 2004; Jordan et al., 2000; Pawson, 2004). Modifications of the posttranslational state, enzymatic activity, or total level of key proteins can act as "molecular signals" that are relayed and interpreted to control cell function. The challenge of identifying which observed molecular signals determine a cell response is complicated because many signaling proteins appear to send mixed or opposing messages. For example, the transcription factor nuclear factor-ICE (NF-rcB) is widely regarded as a prosurvival protein because nuclear relocalization and DNA

binding upregulate expression of apoptosis inhibitors such as c-IAP2, Bcl-xL, and c-FLIP (Karin and Lin, 2002). In response to DNA-damaging agents, however, nuclear NF-KB can promote cell death by recruiting histone deacetylases that silence antia- poptotic genes (Campbell et al., 2004). Molecular signals can not only change their phenotypic meaning but also the relative importance of their message. Tumor cells, for instance, become addicted to chronically activated mitogenic pathways that are used only transiently in normal cells (Weinstein, 2002). Tools that could predict or explain such context-specific roles of molecular signals would be valuable for designing better-targeted therapies against disease (Blume-Jensen and Hunter, 2001; Miller-Jensen et al., 2007).

Many data-driven approaches exist for grouping, separating, or predicting outcomes on the basis of complex quantitative patterns of signaling or gene expression (D'Haeseleer, 2005; Janes and Yaffe, 2006; Noble, 2006). The problem with all of them is that they cannot distinguish molecules that are mechanistically linked to a phenotype from biomarkers that are correlative but not causative (Sawyers, 2008). This difficulty can be avoided by creating models from data sets that consist of molecular signals with recognized but complicated roles in the outcome that is to be predicted (Janes et al., 2005; Miller-Jensen et al., 2007). The drawback is that one's interpretation of such a model is biased toward the recognized roles of the molecular signals and away from more-surprising correlations with phenotype that could indicate new mechanisms. Data-driven models often identify hundreds of correlations in large data sets, making it impractical to perturb each one experimentally. Thus, an additional means for filtering correlation-based hypotheses is greatly needed.

Here, we develop a general approach, called "model-breakpoint analysis," which involves globally perturbing the measurements used to build adata-driven model and then quantifying the loss of model accuracy. We altered signaling-network measurements by manipulating each molecular signal's "dynamic range," defined as the responsiveness of cell outcomes to incremental changes in signal activation. Dynamic range has been under- studied, because signaling networks are typically measured in either their basal (minimum) or hyperstimulated (maximum) states

Leading Edge

Review

Nature, Nurture, or Chance: Stochastic Gene Expression and Its Consequences Arjun Raji and Alexander van Oudenaardenll* 'Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA *Correspondence: [email protected] DO1 10,1016/j.ce11.2008.09.050

Gene expression is a fundamentally stochastic process, with randomness in transcription and translation leading to cell-to-cell variations in mRNA and protein levels. This variation appears in organisms ranging from microbes to metazoans, and its characteristics depend both on the biophysical parameters governing gene expression and on gene network structure. Stochastic gene expression has important consequences for cellular function, being beneficial in some contexts and harmful in others. These situations include the stress response, metabolism, development, the celi cycle, circadian rhythms, and aging.

Introduction Life is a study in contrasts between randomness and deter- minism: from the chaos of biomolecuiar interactions to the precise coordination of development, living organisms are able to resolve these two seemingly contradictory aspects of their internal workings. Scientists often reconcile the stochastic and the deterministic by appealing to the statistics of large numbers, thus diminishing the importance of any one molecule in particular. However, cellular function often involves small numbers of molecules, of which perhaps the most important example is DNA. It is this molecule, usually present in just one or few copies per cell, that gives organisms their unique genetic identity. But what about genetically identical organisms grown in homogenous environments? To what degree are they unique? Increasingly, researchers have found that even genetically identical individuals can be very different and that some of the most striking sources of this variability are random fluctuations in the expression of individual genes. Fundamentally, this is because the expression of a gene involves the discrete and inherently random biochemical reactions involved in the production of mRNAs and proteins, The fact that DNA (and hence the genes encoded therein) is present in very low numbers means that these fluctuations do not just average away but can instead lead to easily detectable differences between otherwise identical cells; in other words, gene expression must be thought of as a stochastic process.

The experimental observation that the levels of gene expression vary from celi to cell is certainly not new. In 1957, Novick and Weiner showed that the production of beta-galactosidase in individual cells was highly variable and random, with induction increasing the proportion of cells expressing the enzyme rather than increasing every cell's expression level equally (Novick and Weiner, 1957). Such early studies were hindered, however, by the lack of reliable single-cell assays of gene expression. One of the first studies to use an expression reporter in single cells to examine the stochastic underpin- nings of expression variability was the pioneering work of KO

et al. (1990). They examined the effect of different doses of glucocorticoid on the expression of a glucocorticoid-respon- slve transgene encoding beta-galactosidase and found that the cell-to-cell variability in the expression of the transgene was surprising large. Moreover, increasing the dose led to an increased frequency of ceils displaying a high level of expression rather than a uniform increase in expression in every cell; that is, dose dependence was a consequence of changing the probability that an individual cell would express the gene at a high level.

Yet, despite the potential biological consequences of random cellular variability (Spudich and Koshland, 1976), several years would pass before theoretical work ignited much of the present interest in stochastic gene expression (McAdams and Arkin, 1997; Arkin et al., 1998). They modeled gene expression using a stochastic formulation of chemical kinetics derived by Gillespie (1977), predicting that in some biologically realistic parameter ranges, protein numbers could fluctuate markedly within individual cells. They then extended their analysis to model the circuit underlying the decision between lysis and lysogeny of the phage lambda, showing that stochastic effects in the expression of key regulators could explain why some cells activated the iytic pathway whereas others followed the lysogenic pathway. The notion that stochastic effects In gene expression could have important biological implications has motivated much research in the field and has only recently been explored experimentally.

Since this early research, the study of stochastic gene expression has blossomed into a rich field, with researchers from a diverse set of backgrounds working on a wide range of problems. The field is also notable for its strong interplay between theory and experiment, with many scientists making significant contributions to both. In this review, we will describe these researchers' efforts to characterize the underlying phe- nomenon through a host of organisms using a variety of experimental and theoretical methods. We will then highlight some recent endeavors trying to tie stochastic gene expression to biological phenomena.

The MatPImatS Site-Specif ic System Organizes the Terminus Region of the E. coli Chromosome into a Macrodomain Romain Mercier,' Marie-Agnes Petit,2 Sophie S ~ h b a t h , ~ Stephane R ~ b i n , ~ Meriem El K a r o ~ i , ~ Frederic Boccard,'r* and Olivier Espelil v*

'Centre de GQn6tique Mol6culaire du CNRS, 91 198 Gif-sur-Yvette, France 'INRA, UR888, Unit6 Bactbries Lactiques et Pathogenes Opportunistes, F-78350 Jouy-en-Josas, France 31NRA, UR1077 MathBmatique, Informatique, et Genome, F-78350 Jouy-en-Josas, France 4AgroParisTech/lNRA, UMR518 Unit6 Mathematlques et lnformatique Appliqu6es, F-75005 Paris, France *Correspondence: [email protected] (F.B.), espeliQcgm.cnrs-gif.fr (O.E.) DO1 10.1 01 6/j.ce11.2008.08.031

SUMMARY

The organization of the Escherichia coli chromosome into insulated macrodomains influences the segregation of sister chromatids and the mobility of chromosomal DNA. Here, we report that organization of the Terminus region (Ter) into a macrodomain relies on the presence of a 13 bp motif called mats repeated 23 times in the 800-kb-long domain. mats sites are the main targets in the E. coli chromosome of a newly identified protein designated MatP. MatP accumulates in the cell as a discrete focus that coloc- alizes with the Ter macrodomain. The effects of MatP inactivation reveal its role as main organizer of the Ter macrodomain: in the absence of MatP, DNA is less compacted, the mobility of markers is increased, and segregation of Ter macrodomairi occurs early in the cell cycle. Our results indicate that a specific organizational system is required iri the Terminus region for bacterial chromosome management during the cell cycle.

INTRODUCTION

The large size of genomes compared to cellular dimensions im- poses extensive compaction of chromosomes compatible with DNA metabolism during replication, transcription, and segregation. Compaction of the chromosome results in the formation of a structure called the nucleoid. Work performed in different bacteria has revealed a number of processes involved in bacterial DNA condensation and organization; they include unrestrained DNA supercoiling, formation of a chromatin-like structure through the interaction of nucleoid-associated proteins (NAPS) with DNA, condensation by structural maintenance of chromosomes (SMC)- like proteins, and macromolecular crowding (Thanbichler and Shapiro, 2006). Cytological analyses have revealed that bacterial circular chromosomes are organized with a specific disposition within growing cells that preserves the linear order of loci in the

DNA (Viollier et al., 2004; Nieisen et al., 2006b; Wang et al., 2006). In E. coli, cytological (Niki et al., 2000) and genetic (Valens et al., 2004) analyses based on long distance DNA interactions revealed a structuring process that spatially insulates large regions of thechromosomecalled macrodomains (MDs). Collisions between DNA sites belonging to different MDs occur at low frequency, and two particular regions, called nonstructured (NS) regions, can interact with both flanking MDs (Vaiens et al., 2004). The Ori MD containing oriC is centered on migS, a centro- mere-like site involved in bipolar positioning of oriC (Yamaichi and Niki, 2004). Opposite the Ori MD, the Ter MD containing the replication terminus is centered on the chromosome dimer resolution site dif. The Ter MD is flanked by the Left and Right MDs, whereas the Ori MD is flanked by the two NS regions. MD organization wasdirectly visualized by analysis of the positioning, the segregation pattern, and the dynamics of markers belonging to various MDs. Markers in MDs showed much lower mobility than markers in NS regions (Esp6li et al., 2008).

Understanding of chromosome segregation has improved considerably in recent years (Thanbichler and Shapiro, 2006). Replication initiates from a single origin, oriC, and progresses bidirectionally to the terminus of replication located opposite the origin. In E. coli, replication initiation occurs In specific replication factories where both replisomes are colocalized. As replication progresses, foci representing the two replisomes follow separate paths (Bates and Kleckner, 2005; Reyes-Lamothe et al., 2008). A colocalization step between sister chromatids of the newly duplicated foci has been reported and appears to vary with growth conditions (Sunako et al., 2001 ; Bates and Kleckner, 2005; Niel- sen et al., 2006a; Adachi et al., 2008; Espeli et al., 2008). This process seems to be more persistent for loci located near the terminus of replication (Li et at., 2003; Bates and Kleckner, 2005; Espbli et al., 2008). The loss of colocalization occurs simultaneously at various positions on the chromosome when much of the chromosome is replicated (Bates and Kleckner, 2005; Espeli et al., 2008).

The mechanisms responsible for structuring the chromosome in MDs are largely unknown. A likely model postulates that internal organization of the different MDs involves recognition of a domain-specific repeated motif by a protein that would isolate

FLOl Is a Variable Green Beard Gene that Drives Biof ilm-like Cooperation in Budding Yeast Scott S r n ~ k a l l a , ~ ~ ~ Marina Caldara,1~8 Nathalie P ~ c h e t , ~ ~ ~ ~ ~ Anne B e a ~ v a i s , ~ Stephanie G ~ a d a g n i n i , ~ Chen Yan,' Marcelo D. Vinces,' An J a n ~ e n , ~ , ~ Marie Christine P r e ~ o s t , ~ Jean-Paul Latge,3 Gerald R. Fink,5 Kevin R. Foster,' and Kevin J. Ver~trepenl . "~** IFAS Center for Systems Biology, Harvard University, Northwest Labs, 52 Oxford Street, Cambridge, MA 02138, USA 2Bioinformatics and Evolutionary Genomics & Quantitative Genomics Groups, Department of Plant Systems Biology, Flanders lnstitute for Biotechnology WIB), and Department of Molecular Genetics, Ghent University, Technologiepark 927 8-9052 Ghent, Belgium 31nstitut Pasteur, Aspergillus Unit 41nstitut Pasteur, Electron Microscopy Platform 25 rue du Docteur Roux, 75724 Paris cedex 15, France 5Whitehead lnstitute for Biomedical Research/M.I.T., Nine Cambridge Center, Cambridge, MA 02142, USA %entre of Microbial and Plant Genetics, Department of Molecular and Microbial Systems, K.U.Leuven, Kasteelpark Arenberg 22, 8-3001 Leuven (Heverlee), Belgium 7Laboratoty for Systems Biology, Flanders lnstitute for Biotechnology (VIE), K.U.Leuven, Heverlee, Belgium 8These authors contributed equally to this work *Correspondence: kevin.verstrepen8biw.kuleuven.be DO1 10.1 016/j.ce11.2008.09.037

SUMMARY

The budding yeast, Saccharomyces cerevisiae, has emerged as an archetype of eukaryotic cell biology. Here we show that S. cerevisiae is also a model for the evolution of cooperative behavior by revisiting flocculation, a self-adherence phenotype lacking in most laboratory strains. Expression of the gene FLOl in the laboratory strain S288C restores flocculation, an altered physiological state, reminiscent of bacterial biofilms. Flocculation protects the FLOl expressing cells from multiple stresses, including antimicrobials and ethanol. Furthermore, FLOI." cells avoid exploitation by nonexpressing flol cells by selfinon-self recognition: FLOl 'keI ls preferentially stick to one another, regardless of genetic relatedness across the rest of the gename. Flocculation, therefore, is driven by one of a few known "green beard genes," which direct cooperation toward other carriers of the same gene. Moreover, FLOI is highly variable among strains both in expression and in sequence, suggesting that flocci~lation in S. cerevisiae is a dynamic, rapidly evolving social trait.

INTRODUCTION

Since Darwin, evolutionary biologists have been troubled by cooperative behavior. Darwin systematically identified the phenomena that were the greatest challenge to his ideas. Coopera- tion was, and remains (Pennisi, 2005), one of these: "lfit couldbe proved that any part o f the structure of any one species had been

formed for the exclusive good of another species, i t would annihilate my theory, for such could not have been produced through naturalselection" (Darwin, 1859). Cooperation is a problem for evolution by natural selection because individuals are predicted to act in a way that maximizes their personal reproduction. Costly behaviors that invest in a common good, therefore, are expected to be disrupted by so-called "cheaters" that save on the cost of cooperation but reap in the benefits of the invest- ment of others. Such cheaters will be fitter than cooperators and take over the population, ultimately resulting In the loss of the cooperative behavior.

Why then do organisms frequently evolve behaviors that help others? For example, honeybee workers labor their whole life without reproducing, birds make alarm calls, and humans often help one another. This fundamental question has received considerable attention over the last 50 years with the development of the field of sociobiology. Following the work of Hamilton (1964), it is now widely accepted that cooperative behaviors evolve because they directly help the actor alongside any recipients, or they help individuals who share more alleles wlth the actor than predicted by chance (genetic relatedness), or both (Dawkins, 1976; Hamilton, 1964; Queller, 1984; West et al., 2006). In extreme cases, therefore, cooperators can successfully transmit their genes by helping another Individual that carries these alleles, as occurs when near-sterile honeybee workers help their mother to reproduce. Typically, it is assumed that the correlation in genotype among individuals is generated by family, as is the case for sister workers in the social insects. However, Hamilton also engaged in a thought experiment, in which he proposed that cooperation is also possible if asingle gene that drives the tendency to cooperate can also preferentially direct cooperation to other carriers of the gene. Such a (hypothetical) gene was later named a "green beard gene" by Dawkins, the green beard

Mistranslation of Membrane Proteins and Two-Component System Activation Trigger Antibiotic-Mediated Cell Death Michael A. Kohanski,*J Daniel J. D ~ y e r , ~ ~ ~ Jamey Wie rzbow~k i ,~ Guillaume Cottarel,* and James J. Collins'~2~3~* 'Howard Hughes Medical Institute 2Department of Biomedical Engineering, Center for BioDynamics, and Center for Advanced Biotechnology Boston University, 44 Cummington Street, Boston, MA 02215, USA %oston University School of Medicine, 715 Albany Street, Boston, MA 021 18, USA 4Department of Math and Statistics, 11 1 Cummington Street, Boston University, Boston, MA 02215, USA *Correspondence: jcollinsObu.edu DO1 10.1 01 6/j.ce11.2008.09.038

SUMMARY

Aminoglycoside antibiotics, such as gentarriicin and kanamycin, directly target the ribosome, yet the mechanisms by which these bactericidal drugs induce cell death are not Fully understood, Recently, oxidative stress has been implicated as one of the mechanisms whereby bactericidal antibiotics kill bacteria. Here, we use systems-level approaches and phenotypic analyses to provide insight into the pathway whereby aminoglycosides ultimately trigger hydroxyl radical formation. We show, by disabling systems that facilitate membrane protein traffic, that mistranslation and misfolding of membrane proteins are central to aminoglycoside-induced oxidative stress and cell death. Signaling through the envelope stress-response two-component system is found to be a key player in this process, and the redox-responsive two-component system is shown to have an associated role. Additionally, we show that these two- component systems play a general role in bactericidal antibiotic-mediated oxidative stress and cell death, expanding our understanding of the common mechanism of killing induced by bactericidal antibiotics.

INTRODUCTION

Aminoglycosides, such as kanamycin and gentamicin, are a powerful class of bactericidal antibiotics that target the 30s subunit of the ribosome (Davis, 1987). The aminoglycoside family of antibiotics falls into the larger aminocyclitol group of 30s ribosome inhibitors, which also includes the bacteriostatic antibiotic spectinomycin. Protein mistranslation through tRNA mis- matching is one of the hallmark phenotypes separating the bactericidal aminoglycosides from the other classes of ribosome inhibitors, including spectinomycin, which are bacteriostatic against Escherichia coli (E. cold (Davis, 1987; Weisblum and Da- vies, 1968). Other phenotypes associated with aminoglycoside

uptake and lethality include changes in membrane potential and permeability (Bryan and Kwan, 1983; Taber et al., 1987).

The lethal mode of action of aminoglycosides is thought to be due to either insertion of misread proteins into the inner mern- brane of E. coli (Bryan and Kwan, 1983; Davis et al., 1986) or ir- reversible uptake of aminoglycosides leading to complete inhibition of ribosome function (Davis, 1987); however, it is still unclear how this latter proposed mechanism contributes to aminoglycoside-mediated killing (Vakulenko and Mobashery, 2003). It has also been suggested that the mechanism of aminoglycoside- induced lethality is a function of more than ribosome inhibition and may be due to inhibition of multiple cellular targets (Han- cock, 1981). In addition, there is a correlation between abnormal protein synthesis, such as mistranslation, and protein carbonyl- ation, a form of oxidative stress (Dukan et al., 2000).

We have proposed that bactericidal antibiotics, including aminoglycosides, induce reactive oxygen species formation, which contributes to drug-mediated cell death (Kohanski et at., 2007). In this proposed model, bactericidal antibiotics perturb metabolism and respiration, leading to increased superoxide production and release or exposure of ferrous iron, which interacts with en- dogenous hydrogen peroxide to form lethal hydroxyl radicals (Dwyer et al., 2007; Kohanski et al., 2007). The sequence of events following the initial drug-target interaction that generate an intracellular environment promoting hydroxyl radical formation remains unknown for the different classes of bactericidal antibiotics.

In this study, we have elucidated the biological events following aminoglycoside interaction with the ribosome that lead to reactive oxygen species formation and contribute to cell death. Through systems-level approaches together with phenotypic and biochemical studies, we show that the protein translocation machinery and the Cpx envelope stress-response two-component system play important roles in oxidative stress-related cell death induced by aminoglycosides. We present evidence that the redox-responsive Arc two-component system is involved in this event, possibly via crosstalk between the Cpx and Arc two-component systems. Our results indicate that aminoglycoside-induced free radical formation is triggered by two-component stress-response system sensing of misfolded proteins in

JOURNAL OF BAC~ERIOI.OOY, NOV. 2008, p. (5983-6905 0021-0193/08/$08.00+0 doi: 10.1 128/JR.00722-08 Copyright O 2008, American Society for Microbiology. All Iiighls lieservccl.

Vol. 190, No, 21

The Origins of 168, W23, and Other Bacillus subtilis Legacy strainsVt D a n i e l R. ~cigler, '" ZoltBn ~ r B g a i , ~ S a b r i n a ~ o d r i g u e z , ' Bastic11 ~ h c v r c u x , ~ Aildrca ~ufllcr, '

Thomas ~ l b c r t , ~ R c t ~ y u a n Bai,'+ Markus wyss,' and John B. pcrkins4 Bacillus Genetic Stock Cerztel; The Ohio State Uizioersity, Cohi~izl~us, Ohio'; DSM Ncitritionlrl Products, IGisercrugst, Switzerlandz;

NilnbleCen Sy,~tems Inc., Madison, Wiscon.sirz7; ~rizll L)SM Anti-InJkctives, DAI Innovcrtion, De@, The ~etherlanrls"

Xcccivcd 21 May 2008/Acceptcd 13 August 2008

Bacillus siibtilk is both a model organism for basic research and an ind~~s t r ia l workhorse, yet there are major gaps in our understancling of tlie genomic heritage and provenance of many widely use(1 strains. We analyzed 17 legacy strains dating to the early years of B, snbtilk genetics. For three-NCIB 36101, I'Y79, and SMY-we performed comparative genonle scque~icitig. For the remainder, we i~sed conventional sequencing to san~ple genon~ic regions expected to show seqnence heterogeneity. Sequence compa~-isons showed that 168, its siblings (122, 160, and 166), and the type strains NCIB 3610 and A'L'CC 6051 are highly siniililr and are liltely descendants of the original Marbnrg strain, although the 168 lineage shows genetic evidence of early don~es- tication. Strains 23, W23, and W23SIt are identical in sequence lo each other but only 94.6% itlcntical to the 0

0

Marburg group in the sequenced regions. Strain 23, tlie probable W23 parent, likely arose froin a contaminant 2 2

in the mutagenesis experiments that producctl168. The remaining strains are all genoniic hybrids, showing one o

or more 'W23 islands" in a 168 genornic backbone. Each traces its origin to transformations of 168 derivatives (D a with DNA from 23 or W23. The common prototrophic lab strain PY79 possesses sirbstantial W23 islands a t its trp and sac loci, along with large deletions that have reduced its genonle 4.3%. SMY, reputed to be the parent 3

3 of 168, is actually a 168-W23 hybrid that likely shares a recent ancestor with PY79. Tl~ese data proviile greater .-.. insight into the genomic history of these 4. subtilis legacy strains. E

-- - ? 4 a

Bacillus subtilis, a modcl orgaliistn for gram-positive back- the "one gene-one cnzyine" niodel previously dcvclopcd for ria, is tlie focus of diverse research ititcrcsts in both acadclnic Neurouyorrr (8) and E,schwichin coli (42) to be extende~l to this 3 and industrial scttings (54, 63, 64). One p r i ~ n a ~ y cmphasis is gram-positive sporc formcr as well. Soon aftc~wards, the Yale 2 sporulation, an archetypical form of cell dcveloprncnt (25, 58, group abandoned its B. srlbtilis studies to pursue other rcscarch 5 66). Of equz~l intcrcst, liowevcr, is the p~odr~ct ion of comnicr- interests (24). Sadly, most of its H. suhtilis collcction, including 4 cially attractivc lcvels of small mctabolitcs and cnzyrncs (46, the wild-typc parent, was subsequently lost. At lcast five 11111- g, 57, 61). Invcstigations with B. siibti1i.s bencfil llom thc case o l tants, however-auxotrophs requiring lhreonine (strain 231, 1 its genctic manipulation (32), the wealth of available physio- nicotinic acid (strain 322), or tlyptophan (strains 160, 166, and 2 logical and biochemical data (64), and thc accessibility 01' a 168)-werc prcsewed and transferred to the l,ossessio~l of 5 < well-annotated gcnoine sequencc (40). Associated t c c h ~ i ~ l o - Cllarlcs Yanofsky. Nearly a decade later, Yatlofsky provided co

gies are leading to a growing understi~nding of the protcome the llmtallts to Jollll Spizizcll (65), wllo in a l ~ l d m a r k publi- 2 % (33,39,74), the transcripton~e (43), the mctabolon~c (Sl), and cation demonstrated that thrce of them-122, l6G, and 168- N

thc rnclabolic [lux pattcrns (26, 41, 59) 01' this organism. 0 could be transformed to prototrophy when cxposcd to DNA

B. subtilis slraills used in virtually all acadclnic r c ~ ~ t l r c l l and from 23 (65). The highly tralisformable straill 168 be- 8 lllatly ii~dustrial processes derive from a single tryptophan- c,l,,c the slll?jccl of follow-Llp stlldics detailing this pllcnonlc-

m

requiring auxotroph, strain 168. Despite its ccnlral importance no11 (6, 79). Slraill 168 was sl~bse(~uelit]y clissemillatec1 aroulld to rcscarch, 0111 knowledge of this strain's gcnonlic hcritagc is the world. ~~~~~~~h~~~ soon developed classical gclletic lllcth- incomplctc. Strain 168 was isolated aftcr B. sril?tilis Marburg ods-and later, rccolnbillant atlcl gcllolllic tccllIlologics--.to was mutagcnized with X-rays by two Yale Ullivcrsity botanists, elucidate the physiology and d~v:veloplnellt of this Paul Uurkholder and Norman Giles (1.5). Their experiments B~ tilc mid-lg70s, so many mutants lxld dcvclopcd from showed tlial for both vcgctative cells and spores oTU. siiblilis, 168 a cclltralized repositoly, the ill^^ ti^ stoclC sublethal doses of UV or X-rays caused high f ~ e q ~ ~ c n c i c s of celltcr (BGSC), was cslablishcd to lllai,ltaitl tiicln (81), ai~xotrophy among suivivors. For illany mutants, the auxotro- strain 168 alld its siblings arc tllc only legacy straitls phic requirement could bc met by a singlc nutrient, allowing sllrviving from earliest years B, ,sLL/,lili,s gcllclics, llow-

cvcr. By the early 1960s, slrain W23 bcgzm appearing regularly in thc literature (20, 69, 77, 78). Its origin has remained a

* Cor~espontling author. Mailing addless: Bacillus GOIIC~IC Stock Center, Thc Ohio State University, 484 W. 12th Avc,, Columbus, OH niystely (34). A fcw researchcrs have explicitly stated, without 43210. phone: (614) 292-5550. F ~ ~ : (614) 202-6773. E-mzlil: zciglcr,l citing cviclcnce, that W23 is derived from Uurltholder ancl Gilcs (@osu.edu. straiii 23 (14). Thc connection is tiot intuitively obvious, how-

'1 Supplemental lllatcrial for this arliclc may be loulld at http://jb cvcr: strain 23 recluircs threolline (49, 651, while W23 is ,)ro- .asm.orgj.

$ Present address: The Johns I-Ioplti~ls University, Scllool of Medi- totrophic and streptomyci~l rcsistailt (69). Although W23 was

cinc, Depa~ lment of Neurosurgery, Baltimore, MD. initially considered a wild-type cquivalcnt lo 168, it was soon " l'ublishcd ;~hcad of print on 22 Augusl 2008. discovcrcd that these strains differed significantly in bacterio-

JOURNAI. 01' BAc~.EKIOLOOY, NOV. 2008, p. 7479-7400 0021 -9193/08/$08.00+0 doi:lO.l128/JB.00823-OX Copy~ight 0 2008, Ao~erican Sociely for Microbiology. All Iiighls Iicscrved.

Vol. 190, No. 22

Escherichia coli Harboring a Natural IncF Conjugative F Plasmid Develops Complex Mature Biofilms by Stimulating Synthesis of

Colanic Acid and curliVt Tliithiwat May alld Satoshi Okabc*

(;rPd~mte Sclzool of Engiizcerii~g Holcltuirlo U~ziversily, Kiln-13, Nisl~i-8, Kita-KLL, Sapporo OGO-8628, .7cy>an

Rcccivetl 12 Junc 2008/Acccptcd 29 August 2008

I t has been shown that Escher.icIiia coli harboring the de~~cpressed IncFI and IncFII conjngative P plasmids form con~plex matirre biofilnls by ulsing their P-pilns connections, whereas a plasmiil-free strain fonns only patchy biofilms. Therefore, in this study we investigated the contribution of a natural IncF conjngative F plasmid to the forn~at iol~ of E. coli biofilms. Unlike the presence of a derepressed F plasmid, tlre presence of a natural IncF F plasmicl pronrolcd biofilni formation by generating the cell-to-cell nrati~rg F pili be~vee11 pairs of F' cells (approxinrately two to four pili per cell) alid by sti~nulating the fo~mation of colanic acid and curli meshwork. Formation of colanic acid and crrrli was reqnired after the initial cleposition of F-pilas connections to generate a three-dimensional mushroom-type biofiln~. In adilition, we demonstrated that the conjugative factor of F plasmicl, rather than a pilus synthesis firnction, was involved in curli production during biofilm fornration, which promoted cell-surface interactions. Ci~rl i played an important role in the nlaturation process. Microarray experinrents were perfonrled to iderrtify the genes involved in curli biusynthesis and regulation. The results suggested that a natural F plasmicl was Inore likely a n external activator that inilireclly promoted carli production via bacterial regulatory systenls (the EnvZ/OnrpR two-component regulators and the RpoS and HN-S global regulators). These data proviclcd new insights into the role of a uatural F plasmid claring the development of E. coli biofilnrs.

Surface-attached microbial communities (biofilms) have distinct morphological and biochemical propelties that distinguish them from frcc-living planlctonic cells (30). 'l'he formation of a three-dimensional (3D) mush~oom-type mature biofilm by Pse~idornorz~rs sp. has bccn dcscribcd as a stepwise process with a t least four clevclopmental stages: (i) initial rcvcrsible attachment of planktonic bacteria to the surface, (ii) a transition to irreve~siblc atlacliment involvillg spccilic adhcsivc I'actors and tlic lo~mat ion ol microcolo- ~iies, (iii) maturation or the microcolonics into a lhin film by production of extracellular polymciic substances in an ad- hesive matrix, and (iv) detachment ol'cclls from the biofilm and return of cclls to the planktonic state (41),

In addition to morphological desc~iplions of biolilm fortna- Lion, there is also increasing intcrcst in the horizontal gene transfcr [hat controls biofilm formation (39). Mobilc gcnclic clclnents mediate horizontal gcnc tr;~nsfer bctwccn bilctcria in natural habitats. These e l c m c ~ ~ t s can be conjugative plasinids, Lransposons, or bacteriophages (24). Although a 1;tboratory strain of Esclzerz'chia coli docs not form cxtcnsivc biofilms spon- t;~ncously, conjugativc plasmid cxprcssion"ptomotes the devcl- opment of thick mature biofilms (18, 35). 11-1 p~cvious sluclies of B. coli harboring a conjugative plasrnid, the w o ~ l t e ~ s identified IncFI and IncFII conjugative plas~nids that potentially pro-

motcd lormation ol' a 3D mushroom-type malurc bioiilm sim- = ilar to a Pseudonronas sp. biofilin (6, 34, 46). Notably, the 5 plasmids previously tested (pOX381h1, an IncPI plasmid 1341: $

r F'traD36, an IncFI plasmid [6]; and Rldrdl9, an IncPII plasmid [46]) ale classified as de~cprcsscd conjugalivc plas~nids $. which constitutively express F pili at all times (15). I-Iowever, co~istitutively exl~rcssed F pili are rccluired for dcvclopnlent of

Z highly organized maturc E. coli biolilms (18) and support bio- 2 film maturation even in the absence oL' Ilagclla, type 1 fimbriac, (D

2 curli, the Ag-43 autotransport protein, or Al-2-mediated cell- to-cell communicalion (34). These findings raise the qacstion of how horizontal gene transfcr can promote the formation of

0 dcnsc biolilms, cspccially via a natural IncF F plasmid, in g which F-pilus forn~ation and thc conjugativc n~cchanism arc reprcsscd. The mechanistic basis of thc putative con~lection between natural Fpi lus formation and bioiil~n inaturation is poorly understood. Furthermore, it is also not Itnow11 how natural IncF F-plasmid-conlaining biofilms arc diffcse~~t in tcrms of ultrastructurc and gelic cxprcssion.

A natural IncF I; plasmid that was origiilally obtained from a host E, coli K-12 strain is iL pioneer conjugative plasmid belonging to the IncFI incompatibility group (15; f-I. Slilmizu, Y. Saitoh, Y. Suda, K. Ueliara, G. Sampci, atid K. Mizobuchi, impublishcd dala). This plasmid is refcrrcd to as the fertility factor (also lcnown as F factor o r scx f~tctor) that allows bac- tcria to produce I; pili necessary for bacterial co~~jugation. All

* Co~rcsponding author. Mailing address: Glotlunlc School of BI- ginccling, Noltkaido University, Kita-13, Nislli-8, Icitn-ICu, Sapporo 060-8628, Jap:ln. Phone and fax: 81-(0)l 1-700-6206. E-mail: sokabc @eng.hokudai.ac.jp.

t Supl~lc~~~enla l ~natcrial for this article {nay be found at htlp://jb .asm.org/.

Published ahcad of print on 12 ScpLcmbcr 2008.

the sequences required for conjugative transmissio~i (20 tru genes) of a natural IncF 1; plasrnid arc prcsent in the 33.3-lrb transfer region (the co~~jugativc factor) of this 99.2-ltb plasmid, while the othcr 65.9 kb (the nonconjugativc factor) contains a number of othcr genetic scqucnccs rcsponsiblc for incompatibility, replication, and other functions (27).

A R T I C L E S Published on Web 10/29/2008

I Jammed Acid-Base Reactions at Interfaces

Julianne M. Gibbs-Davis, Jennifer J. Kruk, Christopher T. Konek, Karl A. Scheidt, and Franz M. Geiger*

Depl,ar/rnent uf' Cl1e17zi~try, N o / ~ t l ~ ~ . v e ~ t i ~ m Uniuers i t )~ , 2145 Slre~.iclan Hocitl, Bv~rrrstorz, IIlirzoL 60208

Received June 6, 2008; E-mail: [email protected]

Abstract: Using nonlinear optics, we show that acid-base chemistry at aqueouslsolid interfaces tracks bulk pH changes at low salt concentrations. In the presence of 10 to 100 mM salt concentrations, however, the interfacial acid-base chemistry remains jammed for hours, until it finally occurs within minutes at a rate that follows the kinetic salt effect. For various alkali halide salts, the delay times increase w~th increasing anion polarizability and extent of cation hydration and lead to massive hysteresis in interfacial acid-base titrations. The resulting impl~cat~ons for pH cycling in these systems are that interfacial systems can spatially and temporally lag bulk acid-base chemistry when the Debye length approaches 1 nm.

Introduction into surfactants at airlwater i~lterfices have been studied as a

Interfacial acid-base processes are ubiquitous in nature. function of electrolyte c~ncentration,~ time delays or hysteresis

Pliysical chemistry tells us that acid-base propelties at intc~ faces was not reported for those highly nlobile two-dimensional

should track acid-base properties in the bulk ~ ~ n d e r equilibriunl systems. Hysteresis, however, was observed in electsoosmotic nieasurenients of silica capil la~ies '~ and calorimetric bull< conditions. However, charge-balancing at the interface7-"as

well as the ~nolecular environments oi the bulk and the equilibriu~n rneasurenlents on colloidal silica,'"he latter of

interfacial species can shift interfacial acid-base equilibria by which can be rationalized using statistical rate t l i e o ~ ~ , ~ ~

lnultiple pK, units from their corresponding bulk solution Here, we apply nonlinear optics to show that interfacial

~ a l u e s . ~ ~ ' ~ - ~ ~ While a molecular-level understanding of these acid-base chetnistry tracks the bulk pI-I at low salt conccntra-

dramatic effects has now cn~erged for equilib~iunn, or steady- tions. In the presence of 10 to 100 niM salt concentrations,

state conditions, their influence on the tinie dependence of howcver, the interfacial acid-base clieniistry remains jilninied

interfacial acid-base reactions is just beginning to he under- for hours, until it filially occurs within ~ninutes at a rate that

stood. For instance, dynaniic measurenients using a surfacc forcc follows the kinetic salt effect. For various alkali halide salts,

balance show that, following progressively longer waiting times the delay tiriles increase with increasing anion polarizability atid

after separation froin contact, mica loses protons to appsoach extent of cation hydration and leacl to massive hysteresis in interfacial acid-base titrations. its equilibrium charge state in water via an activated process

that call be affected by the salt concei~lration in thc surrounding aqueous solutio~i.'~ While pH-sensitive chromophores diluted Section

We use a flow systeni (Figure 1A) described p r e v i o ~ s l y ~ ' ~ ~ ~ to

(1) Ilynes, .I. T. Nrrlrire 1999, 397, 565. tnonitor the interfacial roton on at ion slate in teal time and iri situ

(2) Eigen, M. Arigcw. Clrent., hit. Ed. 1$11g1. 1964, 3, I . using an optical probe that is based or1 the "X'l) ~nelliod",'~ The (3) Pcarson, R. G. J. Am. Cltnn. Soc. 1963, 8.5, 3533. squase root or the measured second harlnonic ge~leration (SHG)"-24 (4) Bo~dwell, F. G. Acc. C ~ I C I I I . Res. 1988, 21, 456, intensity yiclcls the SHG E-field, which depends on the interfacial (5) Stumm, W.; Morgan, J. I. Aqirulic Clrerrri.~fty, Clz~triitnl Iiquil~hriu potential, produced by Ihe interfzaial charges, whose detisity is give11

rrnd I\'clle,s ht Nulurrrl W(tlc,ra, 3rd ed.; Jolill Wdcy 6: SOIIF: New York, by dellsity ol illterracia] S~OH, ~ i o - , alld S ~ O H ~ I 1996.

(6) Raviv, U.; Laural, P.; Klcin, J. Nrrtrrre 2001, 413, 51. groups (Figure IB). For chargecl i~lterl'aces, the second hanilonic

(7) Xiao, X. D.; Vogcl, V.; Shen, Y. R. Cher?~. Pkys. Lett. 1989, 163, generation (SHG) E-field can be rnodeled with the l'arniliar second- 555. order response to which a third-older tern1 is added, accortling to

(8) Uibel, R. 11.; I-Ia~ris, J. M. Ancll. Clrerrt, 2002, 74, 51 12. = x'~'F -'w F - X'"E,E(,~~I~,,.'" The third-order response is give11 (9) Iluang, X.; Kovaleski, J. M.; Wirth, M. J . /Itrol. ClrPtrr 1996, 68,4119, by the llroduct 01 the third-ordcr nonlinear susceptibility, X(3) , the

(10) Bhatlacharyya, I<.; Silz~nann, E. V ; Eisenthal, K. S. ,I. Cltetir. Plrys. 1987, 87, 1442.

(I I) Bain, C. D.; Whitesidcs, G . M. Lnrzgrnuit 1989, 5, 1370. (18) Lamberl, W. J.; Mlddlelon, D. L, Airal. C ~ I C I I I . 1990, 62, 1585. (12) Zhao, X.; Sublahmanyan, S.; Eise~lthal, I<. B. Clrrrr~. P l r v . Lett. 1990, (19) Machcsky, M. L.; Andc~son, M. A. Lo~rgrtiuir 1986, 2, 582,

171, 558. (20) Piasecki, W. Larlgnruir 2003, 19, 9526. (13) Zhao, X.; Ong, S.; Waog, H.; Eisc~~thal, K. 13. Clrerrl. 1'lry.c. LC% 1993, (21) Mil'flin, A. L.; Ge~lh, K. A.; Gcigcr., F. M. J. f' / ly~. Clzcrr~, A 2003,

214, 203. 107, 9620. (14) ITn, K.; Bald, A. J. Lcnrgrrruir 1997, 13, 51 14. (22) Al-Ahadlch, 11. A.; Mifflin, A. L.; Muso~~alih, M. 5.; Geigcr, F, M. ,I. (15) Getsl~cvitz, 0.; Sukc~lik, C . N. J. Arii. Cherrl. S'oc. 2004, 126, 482. I'li)~s. Ctreni. 1j 2005, 109, 16852. (16) Konck, C. T.; Musolmfiti, M. I.; Al-Abatllch, 11. A.; Be1 [in, P. A,; (23) Shell, Y. R. Tile /'rirrci/~le.s of Norrlirzerrr O l ~ t i ~ s ; Johri Wiley & Soris:

Nguyen, S . T.; Geiger, F. M. .I. Ant. C/rer?r Soc. 2004, 126, 11754. New Yolk, 1984. (17) Raviv, U.; Laural, P.; Klcin, J. J. Chorr. I'hys. 2002, 116, 5 167. (24) Eisenlhal, K. B. Cllern. Rev. 1996, 96, 1343.

15444 J. AM. CHEM. SOC. 2008. 130,15444-16447 10.1021/ja804302s CCC: $40.75 O 2008 American Chemical Society

. . C O M M U N I C A T I O N S

Published on Web 10/22/2008

Photoswitchable Nanoparticles Enable High-Resolution Cell Imaging: PULSAR Microscopy

Dehong HU,+ Zhiyuan ~ian,"uwei WU,* Wei wan,* and Alexander D. Q. ~ i * l * Puc$c Nortlzwt.st Nutiorz~~l Lubomtory, Richl~nd, Wnslzington 99352, and Depurtrizerzt o f Cltemistry, Wuslzingtolt

State Univet:sit~~, I'ulli~lcm, Wcrshirzgtorz 99/64

Received July 29, 2008; E-mail: dequanQwsu.edu

Fluorescence imaging has transformed biologici~l sciences and opened a window to reveal biological ~ncchanisills in real timc despite the diffraction limit restricting the optical microscope resolution to -300 nm. Recently, two ultraliigh-rcsoli1[io11 Iluo- rescence n~icroscopic techniques have circunlvcnted the diffraction limit by stochastically photoswitching Iluorophores on and off." " The current selection of photoswitchable Ruorophores for such high- resolution imaging techniques are extrcmcly lirnitcd, ant1 the photoswitching ~nechanisms have not been well btudied. Two methods were reported: photoactivatable tluorcscencc protein (PFP) and a pair oS Cy3 and Cy5 dyes in close proximity. I-Iowevcr, the PFP is much larger than dye molecules, and Cy3-Cy5 pairs sufl'er rrorn stringelit requirements that the environments must be oxygen free. These will limit their celliilx imaging applicatiorls. Thus, 11cw photoswitchable fluorophore development becomes critically i111- portant to apply such ultral~igh-resolutio~l ~nicroscopy broadly."

The ideal photoswitchable probe should integrate high brightness, low molecdar weight, and flexible chemical modifications illto a single molecule so that it will easily adiqlt to the cellular environment. I11 this paper, spiropyran derivatives having the desired characteristics are presented as the exccllet~t photoswitchable fluorophores. Using such a fluorophore, we have achieved photo- actuated unimolecular logical switching attained recpnst~.uction (PULSAR) microscopy, with resolution down to 10--40 nm, far beyond the difl'raction limit. Herein, we observe I~anostructures on glass sur-faces and cellular organelles in fixed cells, which ca1111ot be resolved by co~ive~ltio~~al Ruorescer~ce microscopy. Spiropyrar~ lequires no special treatment to cells and can bc reatlily ruodilied using chemical functional groups for specific ta~geting."~

Unlike the photoactivatable green Iluorescent protein ant1 the Cy3-Cy5 pair in vely close proximity, thc spiroj)yrar~-~~~e~.ocyal~i~~e photoswitching ~nechanism and photochen~ically induced structures are well understood (Figure la). '-I2 T l ~ e spi~.opyrarl Sor11i has negligible visible absorption and conscqucntly no Iluorescence under visible excitation (Figure lb). Its ring-ope~~cd form tnerocyani~~e, however, absorbs strongly at 570 nm illid clnits vivid retl fluorescence (665 nm) in hydrophobic nanoparticlc corcs (Figure I L ~ ) . " ~ UV illumiuation induces the spiropyra1l-to-merocya11i11e conversion and switches on fluorescence; visible photoexcitation of the merocyanine accelerates the back conversion and rcturns the photoswitching dye into the off state. Such clcsircd fli~orescencc photoswitching properties will enable nanometer-l.esolutio~~ PUL- SAR microscopy provided that spiropyran/merocyauine dyes have a high emission rate, long photobleac11ing time, qualified c~nittcd photon count, and efficient photocl~e~nical switching.

Accordingly, the single-molecule photophysics or merocyanine , was studied first. Spin-coating dilute spiropyran ant1 poly (methyl

j R~cilic Northwest Naliont~l Labolxlory. - Washi~lgton Slate University.

10.3 0x1 oe 0 P

- 300 400 600 600 700

0

D, Wavelength (nm)

E mo a r - 100 f - :]viLlbk-yw-q 100 ': 0 ii' 0 iT 0 5 10 15 20 0 2 4 8 1 3

Tlme (s) F Tlme (a)

Figure 1. (A) Spiropyran (spiro-form) and merocyaninc (mero-form) werc polymerized into thc hydrophobic corcs of polyi~icr ~lanol~arlicles, and thus tllcir photocheniical convcrsion occurs within isolated protcctcd nanoparticles. (B) The optici~l absorption (Icft and bottom axcs) of spiro- (black) ant1 mero-form (red) and fluorescence specwa (bottom and right axes) of the spim- (black) and ~nero-fonn (red) arc ploltcd against the wavelength. (C and D) Two typical singlc-n~olccule intensity trajcctotics illusl~~atc that t11e plloloswilching dyc in its eniissivc statc; mcrocyanioc behaves vcry much like norni;~l high-qnantu~ii yicltl dycs such as rllodaiuinc. (E) 'l'lic photobleachi~ig timc histogram reveals that thc photobleaching rcaction follows lirst-order kinetics. (F) The histogram of tolal detected photo~l displays thc number of photons dctcctctl fro~n cach i~iclividual ~nolcculc and such crnission occurrence.

methacrylate) in dichloromethane rcsi~lled in thin films containing isolated photoswitcl~dle dyes. Next, UV illu~nination converted the spiropyran dye to the merocyanine fluorophorc. The detected merocyanine photons produce an emission rate of several ItHz under 0.6 1<W/cm2 excitation, comparable to other high-qua~~(urn yield dyes such as ~~hotlan~ine a11t1 strong enough for single-molecule studies."

Figurc Ic and d display two representative Iluorescence intensity trajectories Sronl 100 pl~otoswitcllable individual merocyaniile molecules, and the trajectories cxhibil typical single ~nolecule signatures such as intensity blinlting and one-step 11hotobleaching. Permanent merocyanine photobleaching is evident. In Figurc I c and d, the ~nolecules were bleached at 10.2 s and at 4.2 s, respectively. Moreover, Figure l e reveals the measured photobleaching ti~nes in a histogra~n, which yields an average photobleaching time of 47 s. The histogram exponential decay suggests that pl~otobleaching obeys lirst-order ltiiletics with a mte constant of -0.02 s- '.

The total number 01 photons that one molecule can emit before photobleaching is an important photophysical property because it determines the ultimate spatial resolution of the PULSAR imaging

10.1021/ja805948u CCC: $40.75 o 2008 American Chemlcal Soclety J. AM. CHEM. SOC. 2008, 130, 15279-15281 m 15279

TIIE JOURNAL OF CI-IEMICAL. PHYSICS 129, 184101 (2008)

Coarse-grained kinetic Monte Carlo models: Complex lattices, multicomponent systems, and homogenization at the stochastic level

Stuart D. Collins, Abhijit ~hatterjee,") and Dionisios G. vlachosb) Del~crrtr~zent of Clremicnl Eligineei Ozg, UII~VPI:Y~~,Y o/'DL'I~IcI~Iz, Nei11czt.k. 11clcr~vc1,a 19716, USA

(Rcceivetl 2 July 2008; acceptetl 30 September 2008; published oliline 10 November 2008)

On-latticc kinctic Monte Citrlo (KMC) sirnulntioris have exte~isively been applied to numcrous systems. Howevcr, thcir applicability is scvcrely lilnitecl to relatively short time and length scalcs. Rcccntly, thc coarse-grainccl MC (CGMC) mcthotl was introducetl to greatly expand the reach of the latticc KMC technique. Jlcrein, wc cxte~id the previous spidial CGMC methods to multicon~polient species andlor site types, The untlerlyiug theory is derived ttncl numerical examplcs arc presented to dcmonstratc the niethod, Furthermore, we introduce the concept ol' homogenization at the stochastic lcvel over a11 site types of a spittially coarse-grained cell. Ilomoge~lization proviclcs s novel coarsening of' the numbcr of' processes, an i1111,ortanL aspect for co~uplex problems plagued by the existence of nurnerous microscopic processes (combinatorial complexity). As cxpectetl, the homogenized CGMC mcthocl outperforms the traditional ICMC niethod on computational cost whilc retaining good accuracy. 0 2008 A~~ie~~icurz I I I S ~ ~ ~ L I ~ P OJ'Phy,sic.s. [DOI: 10.10631J ,30052251

INTRODUCTION

Traclitional kinetic Monte Carlo (ICMC) siniulatio~~s llave en.joyecl inipressive success in the engineering ant1 computational scientific con~munity due to thcir ability to capture, among othcrs, noise, out-of-cquilibrium~ proccsscs, and complex particle interactions, '-"he 1-C methocl allows the simulation of spatial helerogeneo~ls systenls with llanoscopic variation, Howevcl; nlany systcms arc loo coln- putationally demanding for ICMC simulation due to lllultiple reasons. I<MC simulations are capable of capturing roughly 10~--10~lattice points (from lO0X 100 to 1000 X 1000 ntn2, assuming a lattice constant ol' I nm), putting many systems of intercst with conelation lengths at or above the microme- ter range out of reach. Long time scale events, such as pattern formation and aggregation, are also often intractable by I<MC simulation."" Diffusion-controllet1 systcms pose a particular difficulty for ICMC sililulatio~i due to thc hydrocly- n;uniic slowtlown from the overwhellnill numbcr of s~liall iliffi~sion jumps tliitt must bc siniulatcd.lg Cillcnlating long- clista~icc interactions consumes a largc fraction of CPU linle. IJinally, systenis with large reaction networlts involvc too inany indiviclual processes for I<MC to track, store, and search tlir0~1g11. This problem, ternlet1 as co~nbinatorial corn- plexity in this paper, arises in many applications. Exa~nples include biology, duc to thc huge numbcr 01' conformations proteins can ta~ce,"-'~ and epitaxy of mctals, cluc to numcrous dil'fi~sion barriers arising from clilferent local atomic environrne~~ts.'"'~

To cxtc~ld Ilic capabilities of the JCMC method lo longer lime and length scales, tlie coarse-grai~~cd MC (CGMC) method has recently been d e v e ~ o ~ c t l . ~ ' ~ 221"20 l u our all-

proach, neighboring microscopic sites are grouped togethcr into "coarse-grained" (CG) cells it114 a clos~ire is applictl at the stochastic level to resident atonis or molccules (licre itfter ternietl adparticles) to describc thcir distribution in the

1%'",20 In the simplest closure, the local mean-field

(LMF) approximation, adparticles within cells are assu~necl to be well mixed. "~'9~."0 Other closures are exploretl in Ref. 21 and strong i~lteractiolis in Rcf, 22 Tllc atlparticlcs of' each cell are tllen allowed to interact with, reacl with, ancl diffuse to nearby cells.

The CGMC ~ilethod efliciently addresses many of the stated wcal<nesses of tlie tratlitional ICMC method. First, the grouping of microscopic sites simply rcduccs the number of lattice nodes to be indivicl~~ally traclcecl and the number of processes to be simulated. Second, as the size of the CC cells increases, tlie interaction potential length (relative to the CG lattice constant) shrinks, leading to a I-IILIC~ L'Itstcr calculation ol' the coarse interaction polenlial, Third, silnulated tlifl'usion jumps are much largcr, overcoming (in part) the hydrodynamic slowdow~l in cliffusion-co~itrollcd systems.

Prcvious CCMC sirnulatiolls havc l'ocusccl on ~uiiform surli~ces comprised of a single type ol' microscopic sitc with a single typc of aclparticlc. 111 this paper, we extcllcl our prcvious CGMC to an arbitrary number. of site types ancllor adparticle spccics. T,atticc-basctl simulations of multiple site type ancl adparlicle species systems have been performed previously using a MF estimate.'" This extension allows CGMC to be applied to a much wider range of sys-

24 telils of interest, such as catalytic rcaclion systems (where various site types niay represent clifFcrent clellicrlts or latticc positions), cliffi~sio~l on surlbes and in nauoporous ~iiatcrials,"'~'" sane1 biological ~i~nalin~""-~"where sitc types may rcprcscnt distinct arcas of a cell ~nernbrane). In order to

")ll~eseni ;~dd~ess: Thco~eticnl Divis~on, T- 12, MS B268, Los A l r i ~ ~ ~ o s Nn- overcorile the problem of co~ilbinatorial complexity, thc con- (lon;~l l~hora lo~y, Los Alainos, NM 87545.

b)')hutllol w~,olll cOncs,,on~a,cc slloultl I , ~ d c ~ d l C ~ h C ( ~ E ~ e C ~ I O l l l C Illdll. cept of holnogc~lization at thc stochastic lcvel is introcluced. [email protected]~~. el.: 302-83 1-2830. The organization of this paper is as follows. First, the

0021 -9606/2008/129(18)/184101115/$23.00 129, 184101-1 O 2008 American Institute of Physics

Downloacled 20 Nov 2008 to 128.103.149.52. Redistribution subject to AIP license or copyright; see littp:lljcp.aip.org/jcplcopy~'ight.jsp

Simple, robust storage of drops and fluids in a microfluidic device (DO ... http://www.rsc.org.ezp-prod1 .l~~~l.harvard,ed~ddelivery/_ArticleLinki...

Lab on a Chip Journals Lab on a Chip Advance Articles DOI: . 10.1039/b808579j , . . . . . , .. , . . . . . ., ,. , . , , . , ,, . .. ..... .. .. . .... . .

-- -- Lab Chip, 2009 1 DOI: lC!,.1039!!?8085!.?1 1 paper

Simple, robust storage of drops and fluids in a microfluidic

a Hakim Boukellal ", &la SelirnoviC a, Yanwei Jia , Galdcr Cristabal and Seth Fraden * a

"Complex Fluids Group, Martin Fisher School of Physics, Brandeis University, Waltham, MA 02454, USA. E-mail:, fraden@hrandeiis. edzi

b~hodia Asia Pacif c Pte Ltd, Singapore Science Park 11. 11 7586, Singapore

Received 22nd May 2008, Accepted 22nd September 2008

First yublislzed on tlze web 28th October 2008

i

We describe a single rnicrofluidic device and two methods for the passive storage of aqueous drops in a continuous stream of oil without any external control but hydrodynamic flow. Advantages of this device are that it is simple to manufacture, robust under operation, and drops never come into contact with each other, making it unliecessasy to stabilize drops against coalescence. In one method the device can be used to store drops that are created upstream from the storage zone. In the second method the same device can be used to simultaneously create and store drops from a single large continuous fluid stream without resorting to the usual flow focusing or T-junction drop generation processes. Additionally, this device stores all the fluid introduced, including the first amount, with zero waste. Transport of drops in this device depends, however, on whether or not the aqueous drops wet the device walls. Analysis of drop transport in these two cases is presented. Finally, a method for extraction of the drops from the device is also presented, which works best when drops do not wet the walls of the chip.

Introduction

In microfluidics, drops are considered as "micro-reactors", as each drop can be the site of ail independent experiment .'-3 Distinct processing steps include formulation, drop creation and mixing of the contents of the drop. This paper will focus oil related subsequent steps: on-chip drop storage and extraction, which are useful for the class of devices where drop processing occurs on-chip.

Drop formulation and drop creatioii, wliich are conceptually distinct steps, are often intimately related. In this paper drops will always be an aqueous phase while an immiscible oil will be the continuous phase. Popular tecliiiiques for continuous drop generation are "flow focusing"" and the " T-junction"2 where the flow of water solution is cut by a flow of oil. These two methods allow drop production at a frequency as high as tells of ItiloHertz, but have several sliortcoinings. One is tliat formulation is limited to gradual composition variations in successive drops.3 A second shortcolniizg is that there is a start-up time before the flows are steady enough to produce drops of the desired size and frequency and thus it is inevitable tliat some solution is wasted. Mixing the drop's conteilts is achieved by chaotic advectioii when the drop passes around a corner.3 An alteriiate approach to

Non-random segregation of sister chromosomes in Escherichia coli Martin A. white', John I<. ~~ke lenboom' , Manuel A. Lopez-~ernaza', Emily ~ i l s o n ' & David R. F. ~ e a c h "

It has long been k11ow11 that tlie 5' to 3' polarity of DNA sytithesis results in both a lcading and lagging strand at all rcplicatiol~ iol.ksL. Until now, however, there has bee11 no evidcilce tliat leading or lagging strands are spatially orgal~ized in any way within a cell. Here we show that chro~l~osolne segregatioii in Escherichin coli is not ralidonl but is driven i n a manner that results i n tlie leading and lagging strands being addressecl to particular cellular destinations. These destinations are consisterit with the lu~own patterns of chro~nosome s e g r e g a t i ~ n ~ ~ ~ . Our worlc delnonstrates a new level of orga~lization relating to the replication mid segregatiol~ of tlie E. coli chromoso~ne.

I'rokaryotic cclls were long considered to be fcatrucless until recent atlvd~ices ill inlaaing revealed an array of internal structures and sub- - - cellular organizations". One such exa~nplc is the ~ l ~ r c l ~ i t e c t u r e

1. ,i, of 1 co wlicrc the donlains of tlie left and right d1ro1i10soi11e arms o c c ~ ~ p ~ l disti~lct -. Du~ing rcplicdtlon, these do~l la~% arc progressively segrcgatedby an unhlown inechanis<~ that ~,csults in - - - - translational sy111metry of the cl~rornosome arms (known as repli- cho~es) '~~. Co~u~lter-intuitively, to obtain this 11 ansl,ltion,ll symmetl y there must be mirror symlnetry in the scgreg,ition of the leading and laggins st~ands of the two replication forlts (I'ig. In). 'This leads to thice possible situations. Eitller the two I<lgging slrands ofrcylicatioii are positio~led at mid-cell wliilc the lcading str,mds migrate to tlie cell pole&; the two lagging stra~ids rnig~ate to tllc cell poles while the l ead i~~g stra~ids ale positioned at mid-cell; o r a ~andorn co~nl)i~~atioli of these two segregation patterns occurs witlliii the popillatio~l. The

."."Mm"rm"m,/ , I~al~ndrociie 'lead

1 Chromosome b

degradation . ~ " ~ m m ~ " ~ " ~ ~ " " " " " L " ~ , " , ~ m,$-$""""""""'""," ,w ,,<r,,n ",

'lead

Figure 1 I Distinguishing leading and lagging strands. A, C,lrtoon (lemonstrating how mirror sy~~inlmetry or leading and lagging slrands accoimls [or tra~~slatio~~alsy~~~lmnct~y ofsislcr chromosomes. Aa, Scgrcgation of Icatlingslra~~ds to~viuds he ccll polcs and oi'laggi~ig strands towards mid- cell. Ab, Scgrcgation oPlaggi~lgstrands towG~rds the cell polcs a~id of leading slfitnds towards nlid-cell. 0, SbcCD clcaves a DNA hairpin for~l~ed by lhc pali~iclro~l~c 0x1 the lagging strand oTreplication, ;~nd the brolwn chromosome is dcgradcd in a recA- I I I ~ ~ ~ I I L .

specifichypothesis that leading strancls segregate to cell polcs has been proposed'" to explai~i the prcpoliclera~lcc of highly exprcsscd genes on the leading strands of the left (L) and rigllt (I<) replichorcs7. However, no cvide~ice for or against this 11as ~7et been prcscllted.

1'0 distinguish thesc three possibilities, we designed a construct (Supplc~neiitary Fig. 1 ) that would allow us specifically to visu a 1' rze a locus on the right replichorc (kicZ) that was replicated or1 the ledding stla~id of the replication fork (RI,,,t), This is accoillplished by indu- cing SbcCD-~nediatcd palindrome clcavagen in a rccA- i ~ l ~ l t a ~ l t tliat has the palinclro~ile flanked upst~eam hy an array of l e t 0 sites and downstream by an array of lac0 sites. 'l'hcse arrays arc visualized by fluorescence microscopy upon the bi~ldi~lg of l'ctli-yellclw tluor- cscenL protein (YFI') and LacI-cyan fluorescent protein ((;PI'), respectively, and allow us lo follow the cclli~lar position of the chromosomal region contai~iing tlie pdlindro~l~e as it is replicaled and segregated. This is visible as a twin YF1'-C1;P spot, demonstrat- ing the presence of DNA on botl1 sides of the palindron~e. Inciuction of SbcCD results in the specific cleavage of the palindrome that was replicated on the lagging slrand" (Rl,&) by fornlatio~~ of a DNA hairpin". In a recA- mutant the brolten chro~nosome is degl.ddcdl" leaving behind the intact copy of the construct located 011 RI,,,I (Fig. I I)). As long as tllc labelling ~llctllod does 1101 disrupt the Icnown pattern olcl~romoso~ne segregation, then the location of only one of the cl~romosotnc arnis needs to be known for the resL to I)c inferred (Fig. 1A).

The cellular position of this construct was followed during growLll, and the rlumbcr of segrcgatcd copics related to cell length, In the absence of induced double-strancl breaks, 68.8% of exponentially growing cells longer than 1.50 11111 lldd two visihly segregated copies of tlie co~istruct. 111 thcsc cells, the loci were segregated such that one was located at mid-cell and tlle other dt a ccll polc (Fig. 2a). This

Figure 2 1 Visualization of construct. E. coli ccll containing: a, two segrcgatcd copies ol thc construct; b, onc located at micl-ccll; and c, o~ic localetl i ~ t a cell polc. The CFP sig~ialis pseutlocolourctl green, Y FI) magcnta. Overlapping Cl'P and YFP signals appcar ~diitc, Visualization of (lie nucleoid in cclls su1)jected to SbcCD-mediatcd palil~d~ornc clcavagc for 12011lmi11 by DAPI staining (DNA, blue) (d) and mounting on gelatin (DNA, whitc) (e). Scalc bars, I pm (a-c) and 2 pm (d, e).

. . .

'Institute ol Cell Biology, Sc

1248 02008 Macmillan Publishers Limited. All rights reserved

, a Activity motifs reveal principles of timing in 5 transcriptional control of the yeast metabolic network s 0 a, *

Gal chechikl~~, Eugene oh2, Oliver an do', Jonathan weissman2, Aviv ~ e ~ e v ~ & Daphne ~ol ler ' - m C --.

Significant insight about biological networks arises from (! the study of network motifs-overly abundant network

~ubgraphs~~~-but such wiring patterns do not specify when .c( 2 and how potential routes within a cellular network are used.

To address this limitation, we introduce activity motifs, which $ capture patterns in the dynamic use of a network. Using 2 this framework to analyze transcription in Saccharomyces

cerevisiae metabolism, we find that cells use different timing a activity motifs to optimize transcription timing in response to

changing conditions: forward activation to produce metabolic 2 0 compounds efficiently, backward shutoff to rapidly stop 9 production of a detrimental product and synchronized 5 activation for co-production of metabolites required for the 2 same reaction. Measuring protein abundance over a time n ZI course reveals that mRNA timing motifs also occur at the % protein level. Timing motifs significantly overlap with binding 2 activity motifs, where genes in a linear chain have ordered -

binding affinity to a transcription factor, suggesting a mechanism for ordered transcription. Finely timed

8 transcriptional regulation is therefore abundant in yeast N 0 metabolism, optimizing the organism's adaptation to new -

environmental conditions.

ellular processes are mediated through intricate networks of inter- - acting molecules, whose local'" and global"5 topology has been intensively studied. Analysis of network wiring patterns has revealed network motifs-local sets of interaction patterns that occur significantly more often than expected by chance and potentially reflect the functionality of the complex network. However, whereas such network inotifs correspond to the static wiring of the network, neiworlts are used dynamically and adapt to external conditions and internal states in functionally distinct ways.

Such dynamic activity is particularly i~nportant in metabolic processes, which are tightly controlled bascd on the cell's environment.

'Department of Computer Science, Stanford University, Stanford, California 94305, USA. 2Howard Hughes Medical Foundation and Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco, California 94143, USA, 3Departments of Biochemistry and Molecular Pharmacology, University of Massachusetts Medical School, Worcester, Massachusetts 01655,USA. 4~epartment of Biology, Massachusetts Institute of Technology and the Broad lnstitute of MITand Iiarvard, 7 Cambridge Center, Cambridge, Massachusetts 02142, USA. 'Present address: Google Research, 1600 Amphitheater Parkway, Mountain View, California 94043, USA. Correspondence should be addressed to A.R. (aregevQbroad.mit.edu) or D.K. ([email protected]).

Published online 26 October 2008; dol:10.1038/nbt.1499

Upon a change in environmental conditions, a cell may have to rapidly reconfigure its metabolism to produce or degrade compounds to ensure its survival in the new environment. Fluxes through metabolic reactions are controlled by enzymes, the activities and abundances of which are further controlled by post-transcriptional mechanisms. Furthermore, protein abundance is partly determined by transcriptional control6, which modifies the mRNA level of the gene through binding of transcription factors. This complex hierarchy of regulatory mechanisms raises questions about the individual roles and interplay between the different regulation layers. For instance, is transcription regulation tuned to fit the usage patterns of enzymes in the metabolic network? Recent work argues for the predominance of hierarchical control in the metabolic network718. Yet the metabolic network exhibits sigllifica~lt changes in transcript levels in response to environmental perturbations, suggesting the use of transcriptional control. Moreover, recent work on transcriptional control of metabolism identified cases of finer-grained patterns of co-regulation in S. cerevi~iae~~'~. In one specific example in Escherichia coli, the genes in a linear pathway for amino acid biosynthesis were reported to show sequential transcriptional activation ("just-in-time" transcription)".

We developed an analysis framework bascd on the notion of an activity motif (Fig. la) to systematically study the dynamical behavior of such networks. Given a particular network structure (e.g., a linear cascade of enzymes; Fig. lb), an activity motif describes a specific pattern of functional data, such as ordercd timing of activation of the corresponding genes (Fig. lc(i)). Unlike network motifs, which reflect the static wiring of the network (analogous to routes in a road nap)'^'^, activity motifs reflect dymamic and functional patterns associated with their use (analogous to traffic patterns that emerge before, during and after rush hour). Activity motifs can be identified by assessing the enrichment of activity patterns given the network wiring structure.

We applied the frameworlc of activity inotifs to study the dyna~amics oP regulation of gene expression in the metabolic network in S, cerevisiae. Using our systematic analysis, we identified abundant activity motifs involving timed genc expression regulation, recurring in diffcrcnt forms across many conditions. We denlonstrate that the same timing behavior can be conserved in the dynamics of protein abundance, suggesting that timing patterns in inRNA expression can have a direct effect on the timing of metabolic processes. Finally, by studying activity motifs in transcription factor binding affinity, we show that evolution of quantitative transcription factor binding affinities provides a mechanism that can underlie some of this fine- grained control of transcription timing. Overall, the activity motif fralnework allows us to syste~natically investigate three levels of

NATURE UIOTECtiNOLOCY VOLUME 26 NUMBER 11 NOVEMBER 2008 1 2 5 1

A R T I C L E S

2 Protein microarrays with carbon nanotubes as multicolor Raman labels C

% + % Zhuo chen1p4, Scott M ~abakrnan',~, Andrew P ~oodwin', Michael G ~attah', Dan ~aranciang', 5 Xinran wang', Guangyu 2hang1, Xiaolin ~ i l , Zhuang ~ iu ' , Paul J utz2, Kaili liang3, Shoushan an^ & -? Hongjie ~ a i ' E 4 2 The current sensitivity of standard fluorescence-based protein detection limits the use of protein arrays in research and clinical 5 diagnosis. Here, we use functionalized, macromolecular single-walled carbon nanotubes (SWNTs) as multicolor Raman labels for

highly sensitive, multiplexed protein detection in an arrayed format. Unlike fluorescence methods, Raman detection benefits P from the sharp scattering peaks of SWNTs with minimal background interference, affording a high signal-to-noise ratio needed 5 for ultra-sensitive detection. When combined with surface-enhanced Raman scattering substrates, the strong Raman intensity

of SWNT tags affords protein detection sensitivity in sandwich assays down to 1 fM-a three-order-of-magnitude improvement C over most reports of fluorescence-based detection. We use SWNT Raman tags to detect human autoantibodies against proteinase Q 3 3, a biomarker for the autoimmune disease Wegener's granulomatosis, diluted up to lo7-fold in 1% human serum. SWNT 2 Raman tags are not subject to photobleaching or quenching. By conjugating different antibodies to pure 126 and 13c SWNT

isotopes, we demonstrate multiplexed two-color SWNT Raman-based protein detection. s .- r V) .- 2 Several inethods can detect proteins for proteoinic and clinical a diagnostic applications. These include enzyme-linked iillilluilosorbent

assays (ELISAs), fluorcsccncc-bascd protein microarrays', electro- 3 chemistry2, label-free optical methods314, surface-enhanced Raman

scattering (sERs)~-~, microcantilevers8, cluantuill and nano- CO 0 t ~ b e - l ' , ~ ~ or nanowire-based13 field-effect transistors. Among these, 0 N protein m i ~ r o a r r a ~ s ' ~ ~ ' ~ are the most coininon mctliods providing @ high-throughput, multiplexed protein detection for a rangc of appli-

c a t i o n ~ ~ ~ ~ ' ~ ~ ' ~ . Typically, undesirable background interfere~lce or autofluorescence resulting from assay reagents and inaterials limits e$ the sensitivity of protein arrays based on fluorophore tags to - 1 ~ M I I . Increasing the sensitivity of protein detection in arraycd forillat could enhance the capability of this technology for protcoinics research. Moreover, identification of soluble biomarlers for many discases has increased clinical demands for high sensitivity and selectivity in protein detection to facilitate miniillally i~lvasive risk assessment, early-stage disease diagnosis and ~nonitoring of responses to therapeutic int~rventions'~. In addition to improving detectioil sensitivity, protein sensor platforms for diagnosis and research applications would benefit greatly from expanded dynaillic rangcs, which allow more samples to be compared si~nulta~leously with the same standard sct, thereby increasing throughput and reducing the quantity of reagents rcquired.

At least two methodologies under developlnent show particular promise for highly sensitive protein detection in research and clinical applications. Although label-free, nanowire-based t~ansistors'~

demonstrate fe~ntoinolar sensitivity, this sensitivity is liinited to samples in pure water or low-salt solutions, and cannot be achieved in serum or othcr physiological fluids. The second strategy, amplified detection based upon multifunctional nanoparticles, has even greater sensitivity1° but requires multiple reagents and is very time consum- ing. Our methodology, based upon Rainan scattering, has siinplcr rcquirements, may be easily multiplexed and exhibits high sensitivity in clinically relevant samples over the nM to fM range.

The SERS effect providcs the potential for rapid, high-throughput, sensitive protein detection. Although SERS has been applied for immobilized protein detcction by coupling small Raman-active dyes to gold nanoparticlcs functionalized by l i g a n d ~ ~ > ~ , the utility of these sensors is limited by the weak intensities of typical liaman labels, owing to their sinall Raman scattering cross-sectioi~s~~. As a result, sensitivity is not quantitative5, or is limited to the nM range5, which does not compare favorably with fluorescence methods. For high- sensitivity lia~nan sensing with dye molecules7, long acquisition times are needed and inolecules are subject to degradation, resulting from laser radiation.

I11 contrast, SWNTs are ideal labels for SERS-based protein detection. SWNTs have a u n i q ~ ~ e one-dimensional structure and exhibit distinct electrical and spectroscopic properties, includiilg strong and simple resonance Rainan signatures. SWNTs possess enoriilous Raman scattering cross-sections ( - 10W2' cin2 sr-' molecule-'), have simple and tunable spectra and are more stable than other organic Rainan label^^'^^^. Various schemes have been explored to develop

'Department of Chemistry and Laboratory for Advanced Materlals, Stanford University, 333 Campus Drive, Mudd Building, Room 121, Stanford, California 94305, USA. 2School of Medicine, Stanford University, 269 Campus Drive, Stanford, California 94305, USA. 3Department of Physics, Tsinghua-Foxconn Nanotechnology Research Center, Tsinghua University, Beijing 100084, China "These authors contributed equally to the work. Correspondence should be addressed to H.D. ([email protected]).

Received 4 June; accepted 18 September; published online 26 October 2008; doi:10.1038/nbt.l501

NATURE BIOTECHNOLOCY VOLUME 26 NUMBER 11 NOVEMBER 2008 1205

PHYSICAL REVlEW E 78, 041 920 (2008)

Non-Marlcovian noise mediated through anomalous diffusion within ion channels

Sisir ~o~," '"ndrani l ~ i t r a , ~ " and Rodolfo ~ l i n a s ~ ' ; ' ~ l ~ s i c s and Applied Mark~r~zn/ic,s Uliil, Intlinn Sin/i.~licc~l lnslilsrte, 203 Brrrmckpotz fillilk Roacl, Kolkcl/n 700108, Irzdia

2 ~ , n i i ~ & B~havior Prugrar,~, I Iepn~tr i~~nt o/'Plzysics, G~o,gicl S/ate Uizivrrsi/y, 225) P e ~ c h t r r ~ Cetz/er. AVPIILIP, Ailnntn, C;eor<qiiz 3OkZO3, USA

' N ~ W York Ur?iver~i/y SCIIO~I o f Mcdicirze, 530 J3cc.t Aveture, New Rwk, New l?)rh- 10016, USA (licccivcd 20 May 2008; published 28 October 2008)

It is evident fro111 a wide range of cxpctimcntnl findings Lhal ion channcl gating is inherently stochastic. The issue of "memory effccts" (tlifft~sionnl rctardalion due to local changes in watcr viscosity) in ionic flow has been recently addressed using Browni:~~ tlynamics sirlinlations. The results presented indicate such mcmocy cffccts arc ncgligiblc, urlless the dif~usional I~arricr is niuch higher than that of Srcc solutc. In this papa using differential stochaslic ~ncthotls we conclude thal the Markovian properly ol' exponential dwcll ti~iics gives rise to a high barrier, resulling in din'usional memory eITecls Lhat cannot he ignored in dctcrniining ionic flow through channels. Wc have acldlcssed this cluestio~~ using a generalized Langevin cqual~on tlial contains a combination of Markovian and non-Markovian processes with diSfcmnt time scales. This approach afl'oldctl the development ol' an algorithm that tlcscrihes an oscillatory ionic diSC~~sional sequence. The rcs~~lling oscillatory function behavior, with exponcntial decay, was obtained at the wc:tk non-Markovian linlil will1 two distiricl time scales corresponding to Ihc proccsscs of ionic cliI'Susion and drift. This will be analyzed further in firture studics using molecular dynamics simulalions. We propose that the rise oS tinlc scales and memory efl'ccts is rclalcd to diCSercnccs of shear viscosity in thc cyloplasm and extracellular matrix.

DOI: 10.11 03/PIiysRevE.78.04 1920 PACS aumber(s): 87.10.-e, 87.1 6.Xa

I. INTRODUCTION

Ion channels are t~wismcinbrane proteins that iliclude a pore-forming subunit that allows ions to flow between the extracellular and intracell~ilar atid interior of a cell. The open or closed condition oi' such channels inay bc g:~tcd by cillicr the transn~cmbranc electric ficld, via a dipole lno~nenl scnsi- live moiety, or via ligan!l interactions with "chemical sensing" moieties in the channel proteins, Ion channel porcs present a narrow cross section (100 A) and define a path oT low diclcctric constaut across thc membrane. When opcn, the channel pore presents a siathcr specific ion selcclivity filter whcrc thc lines of the electric ficld tc~itl to be conli~lcd to the high dielectric interior of the pore. This paper addresses the problem of pore selectivity.

The continuity rccluircmcnt for the orthogonal cornponcnl of the elcctric displaceme~~t ficld between [he interior of a channel and ~ne~nbrai le is given by E,, ,E~: ,=E, ,E:~~, Since the lipid membrane has a dielectric conslant el,=2, while the clielcct~.ic coilstant of watcr is ~ = 8 0 , it becomes cvitlet~t 1I1al the orthogonal component oS the electric ficld at the membrane pore boundary must be very close to zero. Inclccd, there is o~ily a very slight penctralioll ol' the clcclric Iield inlo the interior of the phospholipid me~~ibrane. 'The situation, therefore, is very similar to the expulsion ol' llie niagnclic ficld by a superconductoc As an cxilmple, in a channcl with a 3 8, radius and a chstlncl of length L=25 A, the barrier is about 6k8T, Although it is quitc large, il shoulcl allow ionic conductivity. This is not too dill'crcnt froun such conditions

where water filled nanopores are introcluccd illto silicon ox- ide films, ~ ~ o l y n ~ e r membranes, elc. [1,2].

Methods ranging from ~nolccular dy~lamics (MD) and Brownian dynamics (BD) (which treat water implicitly as a unifonn diclcctric coi~~inuum) to the mean-field Poisson- Ncl-list Plallck equation (PNP) (which treats both the ions ancl water implicitly as a uniform clielcclric continuu~n) have been used to silnulate the charactcristics ol' ion channel phc- nomena. While MD simulations arc clearly the [nost accu- rate, they are computationally very intensive [3]. Brownian dynamics arc significantly Caster than MD, but becausc of the dielectric tliscontinuitics across lhc various jiilcrl'aces, a new solution of the Poisson equation is required for each configu- ration of the ionic-pore profiles during permeation. The sini- plest approach to study the ionic conduction is based on tlic PNP theory [4,5]. Tliis con~bines the continuity cquation with thc Poisson equation and 011111's atid Fick's laws, PNP is intrinsically nlcan field and is, therefore, bound to fail wllcli ionic correlations become important.

For narrow channels, the cylindrical geometry, conlbi~iecl with the field conlinemcnt, rcsults in a pseudo one- 2

din~ensional potential oi' very long range [6]. U11dcr these coliditio~ls the correlational effccts dominate, and the inean- licld approximation fails [7]. Iiidccd, a rcccnt co~nparisoii between the BD and the PNP showccl [hat the PNP breaks down when the pore raclius is smallel. than aboul two clebye lengths [8,9]. At the nionient, tlicrcfolc, for narrow porcs it appears that a seinicontinuum (implicit solvent) BD siniula- tion is best cornpromised betwcc~l coiilputatiotlal load and accuracy [lo-121. Ii' the interaction potenlial between the ions insidc the clianncl were I<iiown, the siniulalio~i could proceed orclers of magnitude faslcc

Note that though BD sirnulalions ol' ion channels have yielded suitable results, these studies have been coilfilled lo the use of the Lai~gevin equation with Markovian random

920- 1 02008 'I'hc Anlcrican Physical Society

Extensive genomic copy number variation in embryonic stem cells Qi Liang, Nathalie Conte, William C. Skarnes, and Allan Bradley1

Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CBlO ISA, United Kingdom

Edited by Kathryn V. Anderson, Sloan-Kettering Institute, New York, NY, and approved September 12, 2008 (received for review June 10,2008)

lrl Recent analysis of tho human and mouse genomes has revealed gcncratcd d c novo in ES cells in culture, Given that E S cells that highly identical duplicated elements account for 2 5 % of the undergo 30-40 mitotic divisions in vitro before rcsuming thcir sequence content. These elements vary in copy number between normal developmental route into the germ line, we reasoned that individuals. Copy number variations (CNVs) contribute significantly E S cells may accumulate CNVs, which might be transmitted into to genetic differences among individuals and are increasingly the mouse germ line along with targeted alleles, contributing recognized as a causal factor in human diseases with different phenotypic variability to thc allalysis of mutant phenotypes. etiologies. In inbred mouse strains, CNVs have been fixed by inbreeding, but they are highly variable among strains. Within Results strains, de novo germ-line CNVs can occur, leading to interindi- CNVS in ES Cell Clones. W e examined thc genomes of 50 different vidual variation. By analyzing the genome of clonal isolates of E S cell clones for evidence of CNV by comparative genomic mouse ES cells derived from common parental lines, we have hybridization (CGH) against BAC arrays. Givcn that some types uncovered extensive and recurrent CNVs. This variation arises of variation may be incompatible wit11 germ-lint transmission, during mitosis and can be cotransmitted into the mouse germ line and wcwishcd to formulate an unbiased view of the extent of this along with engineered alleles, contributing to genetic variability. variation, wc analyzed clones with confirmed and compromised The frequency and extent of these genomic changes in ES cells germ-line potential. TO maximize the chance of discovering suggests that all somatic tissues in individuals will be mosaics independent events, the clones examined in this study were composed of variants of the zygotic genome. Human ES (hES) cells derived from 3 different parental lines: 2 widely used lines, and derived somatic lineages may be similarly affected, challeng- AB2.2, E14, and the JM8 line that is tllc basis for the EUCOMM ing the concept of a stable somatic genome. and KOMP mutation rcsourcc.

Of 26 clones that could not co~ltribute to the mousc gcrm line, inbred mouse strains I comparative genomic hybridization I BAC arrays trisomies were detected in 7 which involved chromosomes 1, 6,

8, and 11, [supporting information (SI) Fig. S1 and Tables S1 and

C opy number variation (CNV) of DNA scgments in the S2]. In 5 cases, loss of the Y chrornosomc was detected. These human genomc can involvc largc segments of DNA (1-5) types of aneuploidy havc bccn obscrvcd previously, and they

that occur in phenotypically normal individuals and thcsc can be explain the germ-line transmission failure of nearly half of these diseasc-associated (6). Within an inbred mouse strain, CNVs ES clones. In addition to gains and losses of whole chromosomes, also occur among individuals (7,8) and are presumed to arise de 14 germ-line-compromiscd clones exhibited subchromosomal novo by meiotic homologous recombination between nearly changes of 3- to 5-Mb intervals (deletions or duplications) (Fig. identical duplicated sequences and through nonhomologous S2 and Tables S1 S2). These smaller genomic alterations may cnd-joining (9-12). Although the frequency of homologous directly explain thc gcrm-line transmission failure of these recombination is several orders of magnitude greatcr than thc clones, or they may identify breakpoints of more substantial singlc-nucleotide mutation rate, many duplications contain ac- structural rearrangements, such as inversions or translocations

I tive genes; thus, a CNV arising de nova is expected to contribute that cannot be directly detected by CGH. In total, CGH analysis more phenotypic variation (on average) than single-tlucleotide identified genetic changes in 19 of these 26 clones. alterations. Ncxt, wc analyzed the 24 germ-line-compete~lt clones and, as

ES cell lines established from several differcnt strains have expected, none of these clolles exhibited gainior losses of entire bccn thc major route through which thousands of new mutations chromosomes~ However, 7 of these had 3 - to 2 - ~ b deletions have been established in the mouse germ line over the last 2 and/or duplications (Table ~ 2 ) . Subclones derived from all 3 decades. To limit the impact of interstrain variation, the pro- parental cell lines exhibited this type of variation, illcludillg 2 grams generating genome-wide resources of kllockout alleles clonal isolates of the rcccntly derived JM8 cell line, A total of 9

1 (1 3) usc ES cells from a single genetic background, C57BL61N. differcllt C N V ~ wcrc detected in 7 of the 24 germ-linc- I / Thc ~lndcrlying genetic stability of ES cell lines used for thcsc transmittable ES cell clones. ~i~~ of thcsc variants were also

( resources is critical, bccausc the modifiecl ES cell clones used to detected in germ-line-compromiscd clones, ill,jicatillg these establish new germ-line alleles should be genetically identical to specific changes could not explaill germ-lille transmission 1 thc parcntal cell lines. failure of these clones. I-Iowevcr, the number and size of the

Compared with othcr cultured cell lines, ES cells are relatively variants were greater in germ~~~ne~conlprol l l~se~ 1 stablc. Earyotypic variants that arise, such as trisomies or loss of ; the Y chromosome (14, 15), preclude germ-line transmission;

thus, they do not impact genetic studies. However, very little is Author contributions: A.B. designed research; Q.L. and A.B. performed research; N.C. and known about other types of structural variation, such as CNV, W.C.S. contributed newreagentslanalytictoois; Q.L., N.C., W.C.S., and A.B. analyzed data;

which is likely to be compatible with germ-line transmission. and Q.L. and A.B. the paper.

In humans, CNVS have bccn shown to havc a meiotic origin The authors declare no conflict of interest.

(16). Recently, the comparison of monozygotic twins has idcn- This article is a PNAS Direct Submission.

tified CNVS, suggesting thcsc gCnomC altcration~ also OCCUS ITo whom correspondence should be addressed. E-mail: [email protected].

during somatic devclopmcnt (17). Becausc recombination be- his article contains supporting information online at www.pnas.org/cgi/content/fu~~/ tween duplicated sequences occurs at measurable frequencies in 080563BlO5lDCSu~~lemental.

mouse ES cells (18), it would be expected that CNVS would be o 2008 by The National Academy of Sciences of the USA

www.pnas.org/cgi/doi/10.1073/pnas.0805638105 PNAS I November 11,2008 I vol. 105 I no.45 1 17453-17456

The chemistrode: A droplet-based microfluidic device for stimulation and recording with high temporal, spatial, and chemical resolution

PNAS I November4,2008 I vol. 105 I no. 44 1 16843-16848

Delai Chenar1, Wenbin Duasl, Ying Liuarl, Weishan Liua-', Andrey Kuznetsovb, Felipe E. Menderb, Louis H. Philipsonb,

VA and Rustem F. I~rnagi lov~,~

aDepartment of Chemistry and Institute for Biophysical Dynamics, University of Chicago, 929 East 57th Street, Chicago, IL 60637; and bDepartment of Medicine and the Kovler Diabetes Center, University o f Chicago, 5841 South Maryland Avenue, Chicago, IL 60637

Edited by George M. Whitesides, Harvard University, Cambridge, MA, and approved September 15, 2008 (received for review August 19, 2008)

Microelectrodes enable localized electrical stimulation and record- especially of small volcrme, rapidly disperses when transported ing, and they have revolutionized our understanding of the spa- through a tubc by llaminar flow, lcacling to loss of co~lccntration tiotemporal dynamics of systems that generate or respond to ant1 time resolution. Loss of molecules from solution by adsorp- electrical signals. However, such comprehensive understanding of lion to surfaces of tubcs m;ly also occur. Therefore, methods that systems that rely on molecular signals-e.g.. chemical communi- rely on laminar flow to transport molccular signals, such as dircct cation in multicellular neural, developmental, or immune sys- simpling (4), push/pull perfusion (S) , microdialysis (G), and tems-remains elusive because of the inability to deliver. capture, direct microin,jection, have not addressecl this gra~lcl challenge. and interpret complex chemical information. To overcome this 111 contrast to clcctrical signals, molecular signals comprise challenge, we developed the "chemistrode." a plug-based mi- mulliplc, oftcn tml<nown, molecular spccics, rcyuiring ~ h c ability crofluidic device that enables stimulation, recording, and analysis to clclivcr multiple molecular sljecies as stimuli the ability to of molecular signals with high spatial and temporal resolution. analyzc a pulse of response molcculcs by multiple techniques. Stimulation with and recording of pulses as short as 50 ms was Advances in optical imaging technology, new probes and tagging demonstrated. A pair of chemistrodes fabricated by multila~er soft methods, and photo-controllable manipulation have enabled lithography corded independent signals from 2 locations SePa- observation and n~anipulation of inal~y I<nown nlolecular species, rated by 15 pm. Like an electrode, the chemistrode does not need but LIicsc tcchnologics may bc time consu~ning to develop for to be built into an experimental system-it is simply brought into each species and difficult to use for multiple or unknown specics. contact with a chemical or biological substrate, and, instead of " ~ i ~ l ~ ~ ~ a cllip" microfluidics tecllllologies (7-9) can reduce electrical signals, molecular signals are exchanged. Recorded mo- dispersion by minimizing the distance that molecules :Ire trans- lecular signals can be injected with additional reagents and ana- ported by tile illtegratioll of a biological experiment with a lyzed off-line by multiple, independent techniques in parallel (e.g., specific analytical method, I-Iowcvcr, this approach rcquircs the fho~escence correlation spectroscopy. PAALD1-MSf and flu0res- rcdcvclopment and validation of the biologic;tl protocols as well cence microscopy). When recombined, these analyses provide a as tilc milliatllrization alld integration of disparate analytical time-resolved chemical record of a system's response to stirnula- tecllnologies. Rece~lt advances in ~nicrofluidics have used mul- tion, Insulin secretion from a single murine islet of Langerhans was tiPhasc f low to reliably as discrete units measured at a frequency of 0.67 Hz by using the chemistrode.This witllout dilution, cross-conlaminatiol~, or loss of tclnporal res- article characterizes and tests the physical principles that govern olLltioll (IO-.I,Y). the operation of the chemistrode to enable its application to we developed the cllemistrode, a lnicrofluidic platform that probing local dynamics of chemically responsive matter in them- addresses this grand challe~lgc by providi~lg nlolectllar still l~la- istry and biology. tion and recording with high fidelity using plug-based (,12)

n~ultiphase microfluidics /Fig. 1A and supporling information analysis 1 dispersion I f low I microscale I pulse (SI) Fig. SI]. Like the electrode, the chemistrode is simply

brought into contact with thc surface undcr investigation, c.g., a article describes the "chemistrode," a droplet-based cell or tissue, Instead of exchanging electrical sigllals, molecular

rofluidic device for manipulating and observing nlolec- sigllals are delivered by and captured in plugs, aqueous droplets gnals with high spatial and temporal resolution. The nalloliters ill volulne surroullded by a f luorocarboll carrier fluid.

microelectrode, voltage-clamp, and patch-clamp techniques (1) The colnpartmelltalizatim of these lnolecular signals eliminates cn~~blccl s t i~nula t io~~ and recording of electrical activity and dispersioll and loss of salllple due to surface adsorption (18). rcdox-active molecules with high resolution in both spacc and Operation of chemistrode relies on 9 general steps (Fig, 1 time, revolutionizing our understanding of electroactive pro- A andB): (i) preparation of an array of aqueous plugs containing

rom biochemistry to neuroscience (1-3). Most biological all arbitrary secluellce of stimuli (20,21); (ii) delivery of the array i

1 ;

,

i , I

1

I

processes, however, are fundan~entally chcrnical rather than of plugs to a hydrophilic substrate; (iii) coalescence of electrical, rclying on niolccular signals to orchestrate evcnis at thc correct time and location. Elcctrochcmical approaches are widely used, but not all ~ T ~ o ~ c c ~ ~ c s arc ~ ~ C C t ~ ~ C h e ~ l i C ~ ~ ~ y active, Author contributions: D.C., W.D., Y.L., W.L., A.K., L.H.P., and R.F.I. designed research; D.C.,

and soluc elcctrochcmically active lnolccules are difficult to W.D.,Y.L., W.L.,and F.E.M. performed research;D.C.,W.D.,Y.L., W.L.,A.K., L.H.P.,andR.F.i.

selectively in complex mixtures. ~l~~ grand cllallellge analyzed data; and D.C., W.D.. Y.L., W.L., and R.F.1. wrote the paper.

this article addresses is that of dcvising all analogue of the The autllors declare interest. ,

clcctrodc that operates on molccular ratlicr than electrical or This article is a pNA5 Direct Submission.

electroactivc signals. 'D.C., W.D., Y.L., and W.L. contributed equally t o this work.

Why could wc not build such a system with today's technology? 2To whom correspondence should be addressed. E-mail: [email protected].

I Whereas electrical signals trilvcl through wires essentially in- TMS anicle contains supporting information online at www.pnas.org/cgi/conten~f.~

I stailtly ancl with low losses, manipulation and transport of 0807916105lDCSupplemenlal.

I l n ~ k c ~ l e ~ is lnore challenging. First, a pulse of n lo l~cu le~ , o 2008 by The National Academy of Sciences of the USA

Signatures of combinatorial regulation in intrinsic biological noise Aryeh Warmflash and Aaron R. inner'

James Franck Institute, University of Chicago, Chicago, IL 60637

Communicated by Stuart A. Rice, University of Chicago, Chicago, IL, September 19, 2008 (received for review June 10, 2008)

Gene expression is controlled by the action of transcription fac- average concentration of the transcription Iactor inputs by using tors that bind to DNA and influence the rate at which a gene bulk measurcmcnls. Howcver, experinlents performed on single is transcribed. The quantitative mapping between the regulator cells reveal that because transcriptioil factors are often prcscill concentrations and the output of the gene is known as the cis- in low copy numbers, stochastic fluctuations in the concentra- regulatory input function (CRIF). Here, w e show how the CRIF tions 01 these molecules can have inlportallt collscquellces for shapes the form of the joint probability distribution of molecular gene regulation (20-25). Although a dialog between theory and copy numbers of the regulators and the product of a gene. Namely, expcrimcnt has rcvcaled how noise enters the basic processes o l we derive a class of fluctuation-based relations that relate the transcription and translation (26, 27) and propagates through a moments of the distribution to the derivatives of the CRIF. These cascade of genes (23,28,29), the relationship between combina- relations are useful because they enable statistics of naturally aris- torial regulation and stochasticfluctuations has receivedmuchless ing cell-to-cell variations in molecular copy numbers to substitute attention. In particular, it is unclear how the mode of regulation for traditional manipulations for probing regulatory mechanisms. encoded in thc CRIF is rclatcd lo the statistics that character- We demonstrate that these relations can distinguish super- and ize the distribution of protein coilcentralions across a populatio~i subadditive gene regulatory scenarios (molecular analogs of AND of cclls. and OR logic operations) in simulations that faithfully represent Our approach to this area is motivated by the fact that, at bacterial gene expression. Applications and extensions to other equilibrium, the fluctuations of a property of a inaterial call be regulatory scenarios are discussed. related to thc rcsponse to changes in a conjugate control parame-

ter. Awcll-known cxamplc is the relation between the fluctuatioils gene expression I mathematical modeling I noise analysis inference I in the energyof a system and its heat capacity. More generally, the flow cytometry fluctuation-dissipation theorem rclatcs fluctuations about equi-

librium to the relaxation of the system after it has been perturbed from equilibrium (30). General fluctuation-dissipation relations

T ranscription factors regulate the expression of genes by bind- for linear (or linearized) chemical systems that relate the average ing to specific sites 011 the DNA that arc typically spatially values of the chemical concentratio~ls to thc sccond moments can

close to, sometimes even in, sequences that code for proteins. A also be derived (31) have bccll applied to biological systellls group ofsuch binding sites is collectively knowll as a cis-regulatory (29,321. region. When a gene processes the effects of multiple trailscriptioil we use a stochastic model of a regulated by all arbi- factors, one can view it as performil% a comp~llatioll. The inputs trary number of transcriptio~lfactors to show that logic operations are the occupancies of the binding sites in the cis-reg~~latoly region arc most closely related to the derivatives of tile CRIE 111 turn, the and the Output is the rate of transcription. The simplest mcthod 01 derivatives call be related to higher momcllls of tllc dis(ribuLion of describing the relationship between the lra~lscription factors and illput copy numbers ill to molecular sysLcms the output is by usillg booleal1 logic (1, 2). For exalll~le, 2 activa- at equilibrium. These facts suggest meails for inferring rcgulatory tors can regulate agene with AND logic in which boll1 are required synergies from measurements that report on cells. Using or with OR logic in which either transcription factor is sufficie~lt simulations of simple constructs reasollable paralnelers, we for transcription. Knowing which mode of combinatorial regula- demonstrate that the signatures of combillatorial logic be tion a gene employs can be important for dctcrrniiliilg its fullction deteclablc available experilllelltal lnetllods, This is in regulatory nctworks. For cxample, the coherent feed forward useful because proximity in DNA is llot SUfiiCiClli to inler loop, one of the nlOSl Common n10lif~ ill gene regulatory lletwol'lc~ colllbinatorial intcractiolls, and /lley callllot be readily probed by (3), filters noise ill upstream sigllals dilfercntly depending 011 how traditiollal metllods (e.g,, knockouts) or ~ig~l-~~lroLlg~lput expres- one of the participating genes integrates its inputs (4,5). sion assays (e.g., microarray data). Broader implications for the

In general, the output of a gene will llot be binary, and will work are discussed. dcpcnd on the conccnlrations ofthe transcription factors in a complex manncr. Thc notion of logic operalio~ls can be generalized by Results introducing a continuous function that encodes the depe~ldetlce of I, this wc dcvclop a model gene expressioll use the rate of t r~l l~~ript i0nOll the ~0ll~Clltl 'a~ioll~ ofthe illputs. Such it to derive a general expression re]atillg tile distributioll of pro- "cis-rcgulalory input funclions (CRIFs)" (5) have been evaluated tein copy numbers to the derivatives of tile CRIE wc tllell illus- ex~erilnentally for the well-studied operon. The CRIF of the trate tile utility of tllis relatiollship by collsiderillg idealized logic wild type is conlplcx (6, 71, but tllose 01 certain nlutallt operolls gates. Finally, predictions are made for an actual syslcln lo cilable

; reprcscnl molecular analogs of binary logic gates (8). In highcr validation of the lhcory and its assLllllptiolls, 1 organisms, many genes are regulated by a large number o l tran- : scription factors, often with multiple binding sites for cach factor 1 (9-11)' It is illcreasingl~ actiO1l of illall~ regu1a- Author contributions: A w , and A,R,D, &igned A m performed and

tors depends on their molecular context (12-19). Because CRIFs A,W, and A,R,D, wrote the paper. are capable of encoding arbitrarily complex modcs of regulation, The authors declare no conflict of interest. they will likcly bc cven more useful for describing the coillrol of I, ,horn correspondence should be addressed, dinnerQuc~icago.edu. transcriplioll in these s~slemslhan ill s i ln~ler 1Jrokar~oticollcs. his article contains supporting information online at www.pnas.org/cg~/content/full/

To date, the notion of a CRIF has largely been restricted to 0809314105/DCSupplemental.

studying how the average rate of trallscriptioll depends on the @ 2008 by he National Academy of Sciences of the USA

17262-17267 1 PNAS I November11.2008 I vol. 105 1 no. 45 www.pnas.org/cgi/doi/ 10.1073/pnas.0809314105

Analytical distributions for stochastic gene expression vahid'shahrezaeil and Peter S. swain2s3

Centre f o r Non-l inear Dynamics, Department o f Physiology, McGil l University, 3655 Promenade Sir Wi l l i am Osler, Montreal , QC, Canada, H3G 1Y6

Edited b y Charles S. Peskin, N e w York University, N e w York, NY, a n d approved September 5,2008 (received f o r rev iew Apr i l 22, 2008)

Gene expression is significantly stochastic making modeling of By taking the limit of a large ratio of protein to mRNAlifetimes, genetic networks challenging. We present an approximation that wewillstudy the three-stagemodel and a siillpler two-stagcvcrsioi~ allows the calculation of not only the mean and variance, but also whcrc the promolcr is always activc. For this two-slagc modcl, wc thedistribution of protein numbers. We assumethat proteins decay will dcrivc thc prolcin distribution as a f~~nc t ion of lime. We will substantially more slowly than their mRNA and confirm that many derive the steady-staleprotein distribution for the full, three-stage genes satisfy this relation by using high-throughput data from model. We also include expressioils for the corresponding mRNA budding yeast. For a two-stage model of gene expression, with distributions (14,16) in the supporting information (S1)Appenrli.x. transcription and translation as first-order reactions, we calculate the protein distribution for all times greater than several mRNA

A Two-Stage Model o f Gene Expression

lifetimes and thus qualitatively predict the distribution of times We will lirst consider the model of gene expression in Fig. ZA (9). for protein levels to first cross an arbitrary threshold, ~f in addition This model assumes that the promoter is always active and so has the fluctuates between inactive and active states, we can find the two stochastic variables: the number of mRNAs and the number steady-state protein distribution, which can be bimodal if fluctu- of proteins. The probability of having nz mRNAs and n proteins ations of the promoter are slow. We show that our assumptions at timet a master imply that prbtein synthesis occurs in geometrically distributed a p ~ ~ ~ ~ ~ bursts and allows mRNA t o be eliminated from a master equa- --

at - vO(~111-1,1& - P1n,11) -I- ~ 1 ~ ~ ~ l n , l l - ~ - ~ll~, l l )

tion description. In general, we find that protein distributions are asymmetric and may be poorly characterized by their mean and + do[(m + l ) ~ l l z + ~ , l l - mPlll,lL1 variance. Throuqh maximum likelihood methods, our expressions + dl [(n + l)Pm,ll+l - nPl~l,lL1 [I] should therefore allow more quantitative comparisons with experimental data. More generally, we introduce a technique to derive a withv" bcing thc probability per unit timc of transcription, vl bcing

thc probability per unit timc of translation, do being the probabil- simpler, effective dynamics for a stochastic system by eliminating ity per unit time of degradation of an mRNA, and dl bcing thc a fast variable. probability per unit time of degradation of a protein, By defining

the generating function, F(zl,z), by F(zl,z) = C ,,,,,, Z'~Z"P we intrinsic noise I bursts I master equat ion I adiabatic approximat ion

can convert Eq. 1 into a first-order partial differential equation:

G ene expression in both prokaryotes and eukaryotes is inherently stochastic ( 1 4 ) . This stochasticity is both coiltrolled

and exploited by cells and, as such, must be iilcluded in models of genetic networks (5,G). Here we will focus oil describiilg illtrillsic fluctuations, those generated by the randoill timing of iildividual chemical reactions, but extrinsic fluctuations are equally important and arise from the interactio~ls of the system of interest with other stochastic systems in the cell or its ellviroilinent (7,8). Typ- ically, experimental data are compared with predictions of mean l~ehaviors and sometimes with the predicted standard deviation around this mcan, bccausc protciil distributions aic oftcn difficult to derive analylically, even lor models with oilly intrinsic fluctuations.

We will propose a general, although approximate, method for solving thc mastcr cquation for inodcls of geilc exprcssion, Our approach exploits the difference in lifetimes of mRNA and protcin and is valid whcn thc protcin lifctimc is grcatcr thatl the mRNA lifetime. Typically, protei~ls cxist for at least sev- cral mRNA lifctimcs, and protcin fluctualioils arc determined by only time-averaged properties of I~IRNA fluctuations. Following others (7, 9-11), we will use this timc-averaging to simplify the inathcmatical description of stocl~aslic gene expression.

For many organisms, single-cell experiments have shown that gene expression can be described by a three-stage illode1 (3,4,12- 14). The prolnotcr of thc gene of intcrest can transition bctwccil two states (10,15-17), one active and one inactive. Such trailsitioils could be from changes in chromatin structure, froin binding and unbinding 01 protcins involvcd in tianscriptioil (3,4,12), or froill pausing by RNA polynlerase (1 8). Trailscriptioil can oilly occur if thc promotcr rcgion is active. Both trailscriptioil and translation, as well as the degradation of mRNAs and proteins, are usually inodeled as first-order chemical reactions (5).

wherewe have rescaled (19),witha = vo/dl,b = vl/do, y = do/dl, andz =dlt,andwhereu = z l - 1 a n d v = z - 1.

If the protein lifetime is much greater than the mRNA lifetime and y >> 1, Eq. 2 can be solved by using the method of characteristics. Let r measure the distance along a characteristic which starts at z = 0 with u = uo and v = vo for some coilstant u 0 and vo, then Eq. 2 is equivalent to (20)

Consequently direct integration implies r = v = vocT. For 11 >> I, U(V) obeys (SIApperzdix)

Author contributions: V.S. and P.S. designed (esearch, performed research, analyzed data, and wrote the paper.

The authors declare no conflict of interest.

'present address: Department of Mathematics, Imperial College, London SW7 282, UK.

2~resent address: Center for SystemsBiology, University of Edinburgh, Edinburgh EH9 3JU. UK.

3 ~ o whom correspondence should be addressed. E-mail: swainOcnd.mcgill.ca.

This article contains supporting information online at www.pnas.org/cgl/content/fuII/ 08038501051DC5upp1emental.

O 2008 by The National Academy o f Sciences of the USA

17256-17261 1 PNAS I November11.2008 1 vol. 105 1 no. 45 www.pnas.orglcgi/doi110.1073/pnas.0803850105

I Small-scale copy number variation and large-scale changes in gene expression Yuriy Mileykoa-I, Richard I. Johb, and Joshua S. we it^^,^,'

aichool of Biology and Cdool of Physics, Georgia Institute of Technology, Atlanta, GA 30332

b Edited by Simon A. Levin, Princeton University, Princeton, NJ, and approved September 11,2008 (received for review June 30, 2008)

The expression dynamics of interacting genes depends. in part, on mids (23). Morc commonly, rcccnt studics havc attcmptcd to the structure of regulatory networks. Genetic regulatory networks identify statistical relations between CNV and fitness ((4, pro- include an overrepresentation of subgraphs commonly known as tein interactions (24), or combinations of both (25). T o under- network motifs. In this article, we demonstrate that gene copy stand the progression from CNV to changes in phenotype to number is an omnipresent parameter that can dramatically modify changes in fitness, it seems necessary to carefully examine gene the dynamical function o f network motifs. We consider positive regulation itself. The dynamics of a gene regulatory network feedback, bistable feedback, and toggle switch motifs and show depends on network topology, the quantitative nature of feed- that variation in gene copy number, on the order of a single or few backs and interactions between DNA, RNA and protcins, cpi- copies, can lead t o multiple orders of magnitude change in gene genetic modifications of regulatory clcmcnts, and thc biochem- expression and, in some cases, switches in deterministic control. ical state of the intracellular and surrounding environment. Further, small changes in gene copy number for a 3-gene motif Additionally, as we argue here, gene regulatory dynamics can wi th successive inhibition (the "repressilator") can lead t o a qual- also depend sensitively on the copy number of genes and itative switch in system behavior among oscillatory and equilib- promoters. For cxamplc, in synthetically designcd nctworks, rium dynamics. In all cases, the qualitative change in expression is small changes in the copy number of gene regulatory modules due t o the nonlinear nature of transcriptional feedback in which have been shown to lead to qualitative changes in gene expres- duplicated motifs interact via common pools of transcription fac- sion (26, 27). In naturally occurring networks, there may be tors. We are able t o implicitly determine the critical values of copy selection pressures on kinetic parameters such that normally numberwhich lead toqualitativeshifts in system behavior. In some occurring levels of copy number are far from or close to the cases, w e are able t o solve for the sufficient condition for the critical threshold that would lead to a dramatic change in gene existence of a bifurcation in terms of kinetic rates of transcription, expression. translation, binding, and degradation. We discuss the relevance of In this manuscript, we take a quantitative approach to assess our findings t o ongoing efforts t o link copy number variation w i th when small changes in copy number can have a dramatic, cell fate determination by viruses, dynamics of synthetic gene nonlinear effect on gene expression. We study the effect of circuits, and constraints on evolutionary adaptation. changing copy number within a series of small, regulatory

networks commonly referred to as "network motifs'' (28). These gene duplication I gene regulation I network motifs I nonlinear dynamics motifs are network subgraphs shown to be building bloclcs of

complex regulatory networks (29). Increasing the number of

C opy number variation (CNV) is an important and wide- motifs means that multiple networks arc coupled together via a

spread component of within and between population genetic common pool of transcription factors. Changes in the number of

variation. Thc copy nurnbcr of gencs and gcnc fragmcnts varics promotcr sitcs is directly linked to changcs in thc ratc of rcgulatcd rccruitmcnt, which in turn lcads to changcs in trans- significantly over physiological to evolutionary timc scales with

multiple effccts on phenotype. For example, CNV can cause lation and other transcriptional feedbacks (30). We demonstrate that small changes in gene copy number within motifs exhibiting statistically significant changes in conce~ltrations of RNA asso- positive and/or negative feedback can switch the network from ciated with growth rate changes in bacteria (1, 2) and enzyme one alternative steady state to another and switch gene expres- concentrations associated with nutrient intake in humans (3,4), sion to and from an oscillatory state, ThLls, changes ill copy

Thc copy number of viral genomes undergoes d y ~ ~ a ~ n i c a l changes number may act as knobs within a nonlinear dynarnical system durillg m u l t i ~ l c infection of bacteria phages, leading to in much the same way that changes in environmental con~itiolls qualitative changes in gene regulation that may lead to alterna- can drive expression from one steady state to another (29, 31). tive modes of exploitation (5, 6). The duplication of a gene can

I I I facilitate subsequent diversification-a mechanism considered Results

to be a dominant cause of phenotypic innovation (7-10). In extrcmc cascs, wholc-gcnomc duplications havc Icd to lincagc divcrsification within yeast (11). In humans, large-scale deletions and duplications of chromosorncs arc known to causc scvcrc genctic disorders (12, 13) and are imputed in the onset of other diseases including cancer (14, 15). Finally, multiplc studics have demonstrated that CNV in humails is far more extensive than

, I previously bclicvcd, although its impact on phcnotypc is yct to 1 be fully resolved (16-20).

Despite its ubiquity, CNV has been nearly universally over- looked in quantitative models of gene regulation. In those cases

I where quantitative models of CNV have been developed, the primary focus has bccn on changcs in thc copy nurnbcr itsclf, as in the casc of plasmid rnaintenancc (21) and dynamics of transposable elements (22). In somc instances, gcne copy number is integrated into dynamic models of regulatio~l to account for cell-to-cell variability of regulatory clements found on plas-

CNV and Network Motifs. We systematically analyze the depen- dency of 4 network motifs, positive feedback, bistable feedback, toggle switch, and the rcpressilator, on the copy number, N'. The method for analyzing each of these motifs is largely the same, and illustrated in Fig. 1. Although N' is not explicitly prcsent in the mathematical models presented in Fig. 1 A-D, it factors in

Author contributions:Y.M.and J.S.W.designed research; Y.M., R.I.J., and J.S.W. performed research; and Y.M. and J.S.W. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

'To whom correspondence may be addressed. E-mail: yuriy.mileykoOb~ology.gatecli.edu or jsweitzBgatech.edu.

This article contains supporting information online at www.pnas.org/cgilcontent/fulll 08062391051DCSupplemental.

O 2008 by The National Academy of Sciences of the USA

www.pnas.org/cgi/doi/10.1073/pnas.0806239105 PNAS I October28.2008 I vol. 105 I no. 43 1 16659-16664

OPEN 8 ACCESS Freely available online PL"S BIOLOGY

The Pervasive Effects of an Antibiotic on the Human Gut Microbiota, as Revealed by Deep 16s rRNA Sequencing Les ~ethlefsen"~, Sue use^, Mitchell L. sogin3, David A. elma an"^'^* 1 Department of Microbiology and Immunology, Stanford University, Stanford, California, United States of Amerlca, 2 Department of Medicine, Stanford University, Stanford, California, United States of America, 3 Josephine Bay Paul Center for Comparative Molecular Biology and Evolution, Marine Biological Laboratory, Woods Hole, Massachusetts, Unlted States of America, 4 Veterans Affairs Paio Alto Health Care System, Palo Alto, California, United States of America

The human intestinal microbiota is essential to the health of the host and plays a role in nutrition, development, metabolism, pathogen resistance, and regulation of immune responses. Antibiotics may disrupt these coevolved interactions, leading to acute or chronic disease in some individuals. Our understanding of antibiotic-associated disturbance of the microbiota has been limited by the poor sensitivity, inadequate resolution, and significant cost of current research methods. The use of pyrosequencing technology to generate large numbers of 16s rDNA sequence tags circumvents these limitations and has been shown to reveal previously unexplored aspects of the "rare biosphere." We investigated the distal gut bacterial communities of three healthy humans before and after treatment with ciprofloxacin, obtaining more than 7,000 full-length rRNA sequences and over 900,000 pyrosequencing reads from two hypervariable regions of the rRNA gene. A companion paper in PLoS Genetics (see Huse et al., doi: 10.13711 journal.pgen.1000255) shows that the taxonomic information obtained with these methods is concordant. Pyrosequencing of the V6 and V3 variable regions identified 3,300-5,700 taxa that collectively accounted for over 99% of the variable region sequence tags that could be obtained from these samples. Ciprofloxacin treatment influenced the abundance of about a third of the bacterial taxa in the gut, decreasing the taxonomic richness, diversity, and evenness of the community, However, the magnitude of this effect varied among individuals, and some taxa showed interindividual variation in the response to ciprofloxacin. While differences of community composition between individuals were the largest source of variability between samples, we found that two unrelated individuals shared a surprising degree of community similarity. In all three individuals, the taxonomic composition of the community closely resembled its pretreatment state by 4 weeks after the end of treatment, but several taxa failed to recover within 6 months. These pervasive effects of ciprofloxacin on community composition contrast with the reports by participants of normal intestinal function and with prior assumptions of only modest effects of ciprofloxacin on the intestinal microbiota. These observations support the hypothesis of functional redundancy in the human gut microbiota. The rapid return to the pretreatment community composition is indicative of factors promoting community resilience, the nature of which deserves future investigation.

Citation: Dethlefsen L, Huse S, Sogln ML, Relman DA (2008) The pervasive effects of an antibiotic on the human gut microbiota, as revealed by deep 165 rRNA sequencing, PLoS Biol 6(11): e280, doi:l0.1371/Journai.pbio.0060280

Introduction

Specialized microbial communities inhabit the skin, mu- cosal surfaces, and gastrointestinal tract of humans (and other vertebrates) from birth until death, with by far the largest populations in the colon [1,2], Humans rely on their native microbiota for nutrition and resistance to colonization by pathogens [3-61; furthermore, recent discoveries have shown that symbiotic microbes make essential contributions to the development, metabolism, and immune response of the host [7-101. Co-evolved, beneficial, human-microbe interactions can be altered by Inany aspects of a modern lifestyle, including urbanization, global travel, and dietary changes [I], but in particular by antibiotics [I l l . The acute effects of antibiotic treatment on the native gut microbiota range from self-limiting "functio~lal" diarrhea to life-threatening pseu- doinembranous colitis [12,13]. The long-term consequences of such perturbations for the human-microbial symbiosis are more difficult to discern, but chronic conditions such as asthma and atopic disease have been associated with child- hood antibiotic use and an altered intestinal microbiota [14-

161. Because many chemical transformations in the gut are mediated by specific microbial populations [17], with implications for cancer [18,19] and obesity [20,21], among other conditions [22], changes in the composition of the gut microbiota could have important but undiscovered health effects. An approximate return to pretreatment conditions often (but not always) occurs within days or weeks after cessation of antibiotic treatment, as assessed by subjective judgments of bowel f~lnction and characterizations of overall

Academic Editor: Jonathan A, Eisen, Unlverslty of California, Davis, Unlted States of America

Received May 27,2008; Accepted October 6,2008; Published November 18,2008

Copyright: O 2008 Dethlefsen et al.This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Abbreviations: AAD, antibiotic-associated diarrhea; Cp, ciprofloxacin; FDR, false discovery rate; OTU, operational taxonomic unit; PCA, principal component analysis

* To whom correspondence should be addressed, E-mail: [email protected]

:@,', PLoS Biology I www,plosblology.org 0001 November 2008 1 Volume 6 1 Issue I I I e280

SPECIALSECTION

I N T R O U U C T I O N

Getting Your Loops Straight

COMPLICATED BIOCHEMICAL SIGNALING PATHWAYS REGULATE THE FUNCTION OF liviilg cclls. Such regulatory nctworlts oncn have "downstrcam" conlpo~lents that

tion of enormous practical potcntial. -

provide input to comnpone~its that act carlicr in a pathway, crcating fcedback loops. These feedback loops have the potciitial to grcatly altcr thc propertics of a pathway and how it responds to stinluli. To fully understand thesc rcgulatory systems and exploit their vast potential as targets of therapeutic strategies, we need quantitative infomati011 on the flow of signals tlirougll rt pathway and on thc tinling and location ofsignaling cvcnts within cells. Wc liccd to cxplorc the properties that determine, for example, whether a systenl sliows a graded response to a stimulus or turns on and off like a swilcll, how long a patl~way stays activated, whcther its o ~ t - put oscillates, which conlpoucilts call bc pcrturbcd to control the output ofthe system, and so on. The papers assembled in this special issue and in the compa~~ion issuc of Science Signaling highlight rccciit progress in tacltling thcsc challcngcs.

The systems-level approaches tcquircd to u~idc~stand sig~~alingnetworks orten integrate mathe~natical modeling with traditional bioche~llical analysis. Brand- inall and Meyer (p. 390) dcscrlbc thc ways in which fccdback loops allow sophisticated regulatory respoiises, such as adaptable seiisors that respond to changes in the amplitude of an input signal rather than the absolute amount of that signal. Lcwis (p. 399) summar.izcs rccciit exainplcs in which tnodcli~lg approaches allow ncw insights into classical problems in dcvclopmcnt. Spcn~ann's organizer, for example, emits a gradient of sigilaliiig ii~olecules, and recent work explains how the system adjusts whcn an c~nbryo is damagcd, to locrcatc a comnplctc body axis. Tools that allow prccisc noninvasive control and nioniLo~.ing of biocllcmical com- ponents facilitate sorting out how a systcm responds. As Gorostiza and Isacorf cxplain (p. 399, it is now possible not only to samplc thc outl111t ofsignali~lg sys- tc~ns by mo~iitoring the fluorcscct~cc of scpo~ Ler ~i~olcculcs but also to enginccr protcins with light-depe~ldeiit isoincrization swilclies that, when reintsod~~ced into cclls, can bc controlled precisely in time and spacc by cxposure to light.

At Science Sig~zaling (sec www.scic1~cc1i1ag.o~g/ccllsig1~i1lingO8), an original rcsearch paper by Abdi et nl, tales strategies that eilgineers use to uncicrstand the vulnerability of digital circuits and applies them to biological systcms to identify key clenlents of cellular signaling nctworlts. I n Pcrspcctivcs, Dolilrnan discusses how scaffolding molecules call detertni~le the graded or switchlikc response of a patllway. Elsto~i s~~ii~inarizcs how rcsponscs to pulsatile inputs arc used to understand thc dynamic bcliavior ofa signaling system it1 yeast. Chiai~g and M d r outlinc adva~lccs in clicmical approaches provitli~lg deeper insight into signaling mechanisms.

Gcttiiig your loops straight-or undcsstaiiding how a colllplicatcd signaling

. L. BRYAN RAY Science

Signaling /

CONTENTS

Reviews 390 Feedback Loops Shape Cellular

Signals i n Space and Time 0. Brandman and T; Meyer

395 Switches for Remote and Noninvasive Control of Cell Signal t? Gorostiza and E. K lsacaff

399 Signals to Patterns: Space, Time, and Mathematics i n Developmental Biology 1, Lewis

21 October issue ofsclence Signabng

fwww.sciences~gnohng. org).

ing 3 E El

LC

network ~nighlrcspond in ally givcn situatioll-does not colile easily or intuitivcly. But the approaches mentiotlcd hcrc arc clearly opcniiig a laigc area ofinvestiga-

www.sciencemag.org SCIENCE VOL 322 17 OCTOBER 2008 Published byAAAS

able to many different protein classes, potcntially includi~>g cytoplas~natic signaling proteins.

Conclusions Rcnlote and noninvasive mani1,ulation of proteins with light pmvidcs a powe~fil approach for studying and rccngineering sig~ali~lg patl~ways by sclcctivcly establishing a fast and r~ve~sible rcniotc conlrol over specific proteins at sl)ccific locations wihin a cell or organism. TIIC gatest specificity can be obtained fiom light-aclivatcd chemicals that arc lcthc~cd to spccitic proteins. Nat~ually occurring photosensitive proteins have proven to be uscful tools for cxpcrin~cntalion and provide a 1n3jor mo- tivation to seek other naturally lidil-scnsitivc 1x0- tcins. The si~ccess of syntllctic PTL photoswitches introduces a kintalizing prospcct of a gencml method that could be used to confcr control of Inany proleins by light. 'Tllc vc~satilily of the al111roacli co~ucs fium its superticiality. A A L is attached to tllc bland surface of the putein in such a way hat the covalent modification has little impact, cxccpt tliat prcscncc of light of different wavelengtl~s de- tcr~nines whcthcr it prcsc~lls or witlldmws a lig'and to and & O I ~ a ncasby binding site. The ligand can block or activate he site and tlie~~by allcr tlic iilnc- tion of the protein. The site can bc an active sitc or a1 nllosteric i~g~~latory site. This logic should, in principle, be applicable across protein classes.

These optical approaclles to the control of protcin hnction provide an alternative to standard ~nelhods of controlling prolcins through genetic or pha~macological mneans, which tend to bc slow, diifict~lt b Icvelse, and oflen not as speciiic as one

would like. Tile conhast is particularly notablc when one reflects that, in standard pharmacology, high spccificity is associaled with high affinity and tlius with vc~y slow dissociatio~l and a lack of rcvc~sibility. In co~ltlast, spccificity of PTLs ariscs li.om thc close relation between tllc gco~nchy of the PTL and its changc upon isonlcrizalio~l and thc rclativc location oS lllc at(acllmcnt and ligand- binding sitcs. This cvcn pelmils ligands of low aflinity (01 low spccifcity) Lo be used. The extre~nely rapid dissociation latcs of such ligands i111d the subsequent Ihsl reversal of signaling occur. with sa:a(cs similar Lo Iliose of the fastest natuwl biological signals.

References and No tes 1. C. J. Wcijer, Scie~lce 300, 96 (2003). 2. C. Lutz e l a/., Nu t Methods 5, 821 (2008). 3. G. D. Housley, N. P. Raybould, P. R. Thorne, Heat. Res.

119, 1 (1998), 4. 1. Zliao et ol., Biodlemist~y 45, 4915 (2006). 5. Y. Tong e l ol., 1. Gen. Physiol. 117, 103 (2001). 6. S. Shoham, D. H. O'Connor, D. V. Sarkisov, 5. 5. Wang,

Not. Methods 2, 837 (2005). 7. 6. V. Zemelman, N. Nesnas, G. A. Lee, G. Miesenbock,

Proc. Notl. Acod. Sci. U.S.A. 100, 1352 (2003). 8. 5. Q. Lima, G. Miesenbock, Cell 121, 1 4 1 (2005). 9. M. Goeldner, R. Givens, Dynaniic St~tdies i n Biology.

Phototriggers, P i~otuswi tc l~?~ and Caged Biomolecules (Wiley-VCH, Wciliheim, Germany, 2005).

10. E. Bartels, N. H. Wassermann, B. F. Erlanger, Proc. Nail. Acacl. Sci. U.S.A. 68, 1820 (1971).

11. M. Volgraf et ol.,]. Ani. Chem. Soc 129, 260 (2007). 12. G. Nagel et a/., PI-oc, Natl. Acod Sci. U.5.A. 100, 13940

(2003). 13. F. Zliang et a/., Not. Neurosci. 11, 631 (2008). 14. 8. Scliobert, 1. K. Lar~yi, 1. Biol. C11en1. 257, 10306

(1982).

15. A. Bi et ol., Neuron 50, 23 (2006). 16. F. Zhang et a/., Noture 446, 633 (2007). 17. K. Oeisscrotll et a/., I. Neurosci. 26, 10380 (2006), 18. L. Pctrcanu, D. Huber, A. Sobczyk, K. Svoboda,

Not. Neurosci. 10, 663 (2007). 19. G. Nagel et ol., Curr. Biol. 15, 2279 (2005). 20. 5. Scl~roder-Lang et (I/., Not. Methods 4, 39 (2007). 21. M. Bose, D. Groff, I, Xie, El Brustad, P. G. Schultz, I , Am.

Clienl. Sac. 128, 388 (2006). 22. A. Koyer, M. Walko, W. Meijberg, 8. L. Feringa, Science

309, 755 (2005). 23. K. Yoshimura, A. Batiza, C. Kung, Biopi~ys. I]. 80, 2198

(2001). 24. G. A. Woollcy, Act. Chem. Res, 38, 486 (2005). 25. P. Gorostiza, E. Y. Isacoff, Mol. Biosyst. 3, 686 (2007). 26. P. Gorostiza, E. Y. Isacoff, Physiology 23, 238 (2008). 27. 1. Silman, A. Karlin, Science 164, 1420 (1969). 28. H. A. Lester, M. E. Krouse, M. M. Nass, N. H. Wassermann,

B. F. Erlanycr, I. Gen, Pl~ysiol. 75, 207 (1980). 29. D. A. Doyle et ol., Science 280, 69 (1998). 30. R. 0. Blaustcin, P. A. Cole, C. Williams, C. Miller,

Nal. Struct. Biol. 7, 309 (2000). 31. M. Bangllart, K. Borges, E. Isacoff, D. Trauner,

R. H. Kramcr, Not. Neurosci. 7, 1381 (2004). 32. I. I. Chambers, M. R. Bangharl, D. Trauner, R. H. Kramer,

I . Neuropl~ysiol. 96, 2792 (2006). 33. M. Volgraf et a/., Not. Chein. Biol. 2, 47 (2006). 34. P. Gorostiza et ol., Proc. Notl. Acad. Sci. U.S.A. 104,

10865 (2007). 35. 5. Srobota et a/., Neuron 54, 535 (2007). 36. D. 1. Fortin et a/., Nut. Methods 5, 331 (2008). 37. G. C. ElLis-Davies, Nat. Methods 4, 619 (2007). 38. G. Nagel et ol., Bioche~n. Soc. Troia, 33, 863 (2005). 39. P.G. is supported by tlie Human Frontier Science

Program Organization through a Career Development Award and by the European Researcli Council through a Starting Grant. This work was supported by the NIH Nanomedicine Development Center for the Optical Control of Biological Function (grant 5PNZEY018241).

REVIEW

From Signals to Patterns: Space, Time, and Mathematics in Developmental Biology lulian Lewis

We now have a wealth of information about the tnolecular signals that act on cells in embryos, but how do the control systems based on these signals generate pattern and govern the timing of developmental events? Here, I discuss four examples to show how matlie~natical modeling and quantitative experimentation can give some useful answers. The examples concern the Bicoid gradient i n the early Drosophila embryo, the dorsoventral patterning of a frog embryo by bone morphogenetic protein signals, the auxin-mediated patterning of plant meristems, and the Notch-dependent somite segmentation clock.

D evclopmental biologists are preoccupied those elsewhere to adopt anoll~cr'? What controls with patterns in space and pattcms in thc timing of thcsc paLtcl.ning events as thc oga- time. What causcs cells in OIIC par1 of an nisrn dcvclops fiom the egg? What halts growth

ctiibryo regularly to adopt onc cha~ictcs, and when a slructulc has reached its proper size? -- . . - . Eady discussions of devclo~~mcntal patterning

vertebrate Development ~ ~ b ~ ~ ~ t ~ ~ ~ , cancer ~~~~~~~t~ UK considered, in abstract (einls and in tolal igno~;~nce London Research Institute, London WC2A 3PX, UK. of the ~no~cculcs involved, how signals exchanged

betweell cells might drive Ibrn~alion of tlie pattecns we obscwc. AIU~ough hey laclced a conclete 1110- lecular basis, seve~al of these speculative tllcories turn out to lii~ve been reniarkiibly prescient. More- ova; right 01. wrong, thcy drivc homc a gcncral Icsson. To cxl>Iain liow cmnblyos generate their spatial pattc~ns and their Lcni~~oral prog~atns, wc need morc than lists of the molecules i~lvolvcd and morc than simple cartoon diagrams of' the q~~alitativc control relationships belwcen tllcm. 'To ~nal<e sense of tlle control circuil~y, qualitative dala and unaidcd intuition arc not c110ugh. Evcn for the sinll~lcsl cascs, we need mathematics and detailcd mcasurclncnts, just as wc nccd math- c~natics and measurctnc~ils to explain thc olbits of the plane& or the swing of a pcnduhn.

Thc necessity adscs cspccially bccausc of thc filnda~nental pa11 that fcedbacli plays in dcvclop- ~ncnlal control processes, as tliscusscd elsewhere in this issue of Science by l3landman 2uid Meyer (1). Posilivc fcedback can give a systcni a flip/llop choice bctwccn altcrnativc steady states; it can endow a systcln with cndiuing rncmoly of its cx- posurc to past signals; it can gcnciatc idiomoge- ncily in a systcrn that sta11s oul spatially u~iifor~n. Ncgalive fcedback can snlootll out ir~~gillaritics; it can enable a systcm to respond to a signal more

\

www.sciencemag.org SCIENCE VOL 322 17 OCTOBER 2008 399

single-cell chemical lysis method for analyses of … · single-cell chemical lysis method for...

Documents